PA RT F O U R

HUMAN PERFORMANCE

269

EFFICIENCY OF SELECTION 270Failures of Selectivity 270Selection by Location and Other Features 276

PREATTENTIVE AND ATTENTIVE PROCESSING 280Distributed Attention Paradigms 280Inattention Paradigms: Dual-Task Experiments 282

Further Explorations of Preattentive Processing 282Attention, Types and Tokens 285

CLOSING COMMENTS 287REFERENCES 288

We live in a sea of information. The amount of informationavailable to our senses vastly exceeds the information-processing capacity of our brains. How we deal with thisoverload is the topic of this chapter—attention. Consideryour experience as you read this page. You are focused onjust a word or two at a time. The rest of the page is availablebut is not being actively processed at this time. Indeed, quiteapart from the other words on the page, there are many stim-uli impinging on you that you are probably not aware of, suchas the pressure of your chair against your back. Of course, assoon as that pressure is mentioned you probably shifted yourattention to that source of stimulation, at which point youmost likely stopped reading briefly. Some external stimuli donot need to be pointed out to you in order for you to becomeaware of them—for example, a mosquito buzzing aroundyour face or the backfire of a car outside your window. Thesesimple observations point to the selectivity of attention, itsability to shift quickly from one stimulus or train of thoughtto another, the difficulty we have in attending to more thanone thing at a time, and the ability of some stimuli to capture

attention. These are all important aspects of the topic of at-tention that will be explored in this chapter.

Perhaps the most fundamental point about attention is itsselectivity. Attention permits us to play an active role in ourinteraction with the world; we are not simply passive recipi-ents of stimuli. A great deal of the theoretical focus of re-search on attention has been concerned with how we come toselect some information while ignoring the rest. Work in theyears immediately following World War II led to the devel-opment of a theory that holds that information is filtered at anearly stage in perceptual processing (Broadbent, 1958). Ac-cording to this approach, there is a bottleneck in the sequenceof processing stages involved in perception. Whereas physi-cal properties such as color or spatial position can be ex-tracted in parallel with no capacity limitations, further per-ceptual analysis (e.g., identification) can be performed onlyon selected information. Thus, unattended stimuli, which arefiltered out as a result of attentional selection, are not fullyperceived.

Subsequent research was soon to call filter theory intoquestion. One striking result comes from Moray’s (1959)studies using the dichotic listening paradigm, in which head-phones are used to present separate messages to the two ears.The subject is instructed to shadow one message (i.e., repeatit back as it is spoken); the other message is unattended.Ordinarily, there is little awareness of the contents of theunattended message (e.g., Cherry, 1953). However, Morayfound that when a message in the unattended ear ispreceded by the subject’s own name, the likelihood of report-ing the unattended message is increased. This suggests that

CHAPTER 10

Attention

HOWARD EGETH AND DOMINIQUE LAMY

The preparation of this chapter was supported in part by grants toHoward Egeth from NIMH (R01MH57388) and the FAA (2001-G-020). The authors would like to thank Alice Healy, Andy Leber,Melanie Palomares, Robert Proctor, Irving Weiner, and Steve Yantisfor helpful comments, Robert Rauschenberger for preparation of thefigures, and Terri Dannettel for technical help in the preparation ofthe manuscript.

270 Attention

the unattended message had not been entirely excluded fromfurther analysis. Treisman (1960) proposed a modification ofthe filter model that was designed to handle this problem. Sheassumed the existence of a filter that attenuated the informa-tional content of an unattended input without eliminating itentirely. This model was capable of predicting the occasionalintrusion of meaningful material from an unattended mes-sage. However, the discrepant data that led Treisman to pro-pose an attenuator instead of an all-or-none filter led othertheorists in quite another direction. Thus, according to thelate-selection view (e.g., Deutsch & Deutsch, 1963), percep-tual processing operates in parallel and selection occurs afterperceptual processing is complete (e.g., after identification),with capacity limitations arising only from later, response-related processes.

After nearly three decades of intensive research and de-bate, recent reviews have suggested that the apparent contro-versy between the two views may stem from the fact that theempirical data in support of each of them has typically beendrawn from different paradigms.

For instance, Yantis and Johnston (1990) noted that evi-dence favoring the existence of late selection was typicallyobtained with divided-attention paradigms (e.g., Duncan,1980; Miller, 1982). These findings showed only that therecan be selection after identification, rather than entailing thatselection must occur after identification. Yantis and Johnston(1990) set out to determine whether early selection is at allpossible. By creating optimal conditions for the focusing ofattention, they showed that subjects were able to ignore irrel-evant distractors, thus demonstrating the perfect selectivitythat is diagnostic of early selection. Yantis and Johnstonproposed a hybrid model with a flexible locus for visualselection—namely, an early locus when the task involves fil-tering out irrelevant objects, and a late locus, after identifica-tion, when the task requires processing multiple objects.

Kahneman and Treisman (1984) noted that whereas theearly-selection approach initially gained the lion’s share ofempirical support (e.g., von Wright, 1968), later studies pre-sented mounting evidence in favor of the late-selection view(e.g., Duncan, 1980). They attributed this dichotomy to achange in paradigm that took place in the field of attentionbeginning in the late 1970s. Specifically, early studies usedthe filtering paradigm, in which subjects are typically over-loaded with relevant and irrelevant stimuli and required toperform a complex task. Later studies used the selective-setparadigm, in which subjects are typically presented with fewstimuli and required to perform a simple task. Thus, based onthe observation that the conditions prevailing in the two typesof study are very different, Kahneman and Treisman cau-tioned against any generalization across these paradigms.

Lavie and Tsal (1994) elaborated on this idea by proposingthat perceptual load may determine the locus of selection.They showed that early selection is possible only under con-ditions of high perceptual load (viz., when the task at handis demanding or when the number of different objects in thedisplay is large), whereas results typical of late selection areobtained under conditions of low perceptual load. In otherwords, when the task is not demanding, the spare capacitythat is unused by that task is automatically diverted to theprocessing of irrelevant stimuli.

The idea of a fixed locus of selection (whether early orlate) implies a distinction between a preattentive stage, inwhich all information receives a preliminary but superficialanalysis, followed by an attentive stage, in which only se-lected parts of the information receive further processing(Neisser, 1967). The preattentive stage has been further char-acterized as being automatic (i.e., triggered by external stim-ulation), spatially parallel, and unlimited in capacity, whereasthe attentive stage is controlled (i.e., guided by the observer’sgoals and intentions), spatially restricted to a limited region,and limited in capacity. Within this framework, an importantquestion becomes, To what extent are stimuli processed dur-ing the preattentive stage?

One implication of the proposed resolutions of the early-versus-late debate (Kahneman & Treisman, 1984; Lavie &Tsal, 1994; Yantis & Johnston, 1990) is that one cannot drawinferences from findings concerning the locus of selection tothe question of how extensive preattentive processing is. Thatis, how efficient selection can be and what is accomplishedduring the preattentive stage are separate issues. For instance,the idea of a flexible locus of selection advanced by Yantisand Johnston (1990) implies that the level at which selectioncan be accomplished does not reveal intrinsic capacity limi-tations but depends only on task demands, and thus doesnot tell anything about preattentive processing. Similarly,the finding that perceptual load is a major determinant ofselection efficiency (Lavie & Tsal, 1994) makes a usefulmethodological contribution, because it shows that a failureof selectivity does not reveal how extensively unattended ob-jects are processed, but may instead reflect the mandatory al-location of unused attentional resources to irrelevant objects.

EFFICIENCY OF SELECTION

Failures of Selectivity

Various factors affect the efficiency of attentional selection.As was mentioned earlier, Lavie and Tsal (1994) proposedthat low perceptual load may impair selectivity because spare

Efficiency of Selection 271

attentional resources are automatically allocated to irrele-vant distractors. Similarly, grouping between target and dis-tractors may impair attentional selectivity. Another case ofselectivity failure is evident in the ability of certain known-to-be-irrelevant stimuli to capture attention automatically.

Effects of Grouping on Selection

The principles of perceptual organization articulated by theGestalt psychologists at the beginning of the last century(e.g., proximity, similarity, good continuation) correlate cer-tain stimulus characteristics with the tendency to perceivecertain parts of the visual field as belonging together—that is,as forming the same perceptual object. (For a fuller discus-sion of the Gestalt principles see the chapter by Palmer in thisvolume.) Kahneman and Henik (1981) considered the possi-bility that such grouping principles may impose strong con-straints on visual selection, with attention selecting wholeobjects rather than unparsed regions of space. Beginning inthe early 1980s, this object-based view of selection hasgained increased empirical support from a variety of experi-mental paradigms.

Rock and Gutman (1981) showed object-specific atten-tional benefits in an early study. Subjects were presented witha sequence of 10 stimuli, each of which consisted of twooverlapping outline drawings of novel shapes, one drawn inred and one in green. Thus, in each of the overlapping pairs,the two shapes occupied essentially the same overall locationin space. Subjects were required to make aesthetic judgmentsconcerning only those stimuli in one specific color (e.g., thered stimuli). At the end of the sequence, they were given asurprise recognition test. Subjects were much more likely toreport attended items (those about which they had renderedaesthetic judgments) as old than to report unattended items asold. In fact, unattended items were as likely to be recognizedas were new items.

This finding shows that attention can be directed to one oftwo spatially overlapping items. Note, however, that object-based selection was required by the task, which leaves openthe possibility that object-based selection may not be manda-tory. Moreover, the fact that the unattended stimulus wasnot recognized does not necessarily entail that it was notperceived; in particular, it may have been forgotten during theinterval between presentation and the recognition test.

In a later article, Duncan (1984) explicitly laid out the dis-tinction between space-based and object-based views of at-tention and tested them with a perceptual version of the Rockand Gutman (1981) memory task. In Duncan’s study, object-based selection was no more task relevant than space-basedselection. Subjects were presented with displays containing

two objects: an outline box and a line that was struck throughthe box (see Figure 10.1). The box was either short or tall,and had a gap on either its left or right side. The line wasdashed or dotted and was slanted either to the right or to theleft. Subjects were found to judge two properties of the sameobject as readily as one property. However, there was a decre-ment in performance when they had to judge two propertiesbelonging to two different objects. These results showed adifficulty in dividing attention between objects that could notbe accounted for by spatial factors, because the objects weresuperimposed in the same spatial region.

This very influential study has generated a whole body ofresearch concerned with the issue of object-based selection, al-though it has been criticized by several authors (e.g., Baylis &Driver, 1993). Later studies where the problems associatedwith Duncan’s study were usually overcome also demon-strated a cost in dividing attention between two objects(e.g., Baylis & Driver, 1993; but see Davis, Driver, Pavani, &Shepherd, 2000, for a spatial interpretation of object-basedeffects obtained using divided attention tasks).

Recently, Watson and Kramer (1999) added an importantcontribution to this line of research by attempting to specify apriori the stimulus characteristics that define the objects uponwhich selection takes place. They proposed a frameworkthat allows one to predict whether object-based effects will befound, depending on stimulus characteristics. Borrowing fromPalmer and Rock’s (1994) theory of perceptual organization,they distinguished among three hierarchically organized lev-els of representation: (a) single, uniformly connected (UC)regions, defined as connected regions with uniform visualproperties such as color or texture; (b) grouped-UC regions,which are larger representations made up of multiple single-UC regions grouped on the basis of Gestalt principles; and(c) parsed-UC regions, which are smaller representations seg-regated by parsing single-UC regions at points of concavity

Figure 10.1 Two sample stimuli used in the study of object-based atten-tion. Each stimulus consisted of two objects (a box and a line passing throughthe box). See text for further details. Source: Reprinted from Duncan(1984), with permission from the American Psychological Association.

272 Attention

(e.g., the pinched middle of an hourglass is such a region ofconcavity; it permits parsing the hourglass into its two mainparts, the upper and lower chambers).

They used complex familiar objects (pairs of wrenches)and had subjects identify whether one or two predefinedtarget properties were present in these objects (see Fig-ure 10.2). They examined under which conditions object-based effects (i.e., a performance cost for trials in which twotargets belong to different wrenches rather than to the samewrench) could be obtained for each of the three representa-tional levels. They found that (a) object-based effects are ob-tained when the to-be-judged object parts belong to the samesingle-UC region, but not when they are separate single-UCregions, and concluded that the default level at which selec-tion occurs is the single-UC level; and (b) selection mayoccur at the grouped-UC level when it is beneficial to per-forming the task or when this level has been primed.

The finding that it is easier to divide attention between twoproperties when these belong to the same object suggests thatperceptual organization affects the distribution of attention.Another empirical strategy used to reveal these effects is toshow that subjects are unable to ignore distractors when theseare grouped with the to-be-attended target (e.g., Banks &Prinzmetal, 1976). Other studies following this line of rea-soning used the Eriksen response competition paradigm(Eriksen & Hoffman, 1973), where the presence of distractorsflanking the target and associated with the wrong response is

shown to slow choice reaction to the target (see the chapter byProctor and Vu in this volume). They demonstrated that dis-tractors grouped with the target (e.g., by common color orcontour) slow response more than do distractors that are notgrouped with it, even when target-distractor distance is thesame in the two conditions (Kramer & Jacobson, 1991).

Perhaps the strongest support for the idea that attention se-lects perceptual groups rather than unparsed locations wasprovided by Egly, Driver, and Rafal’s (1994) spatial cueingstudy. Subjects had to detect a luminance change at one of thefour ends of two outline rectangles (see Figure 10.3). Oneend was precued. On valid-cue trials, the target appeared atthe cued end of the cued rectangle, whereas on invalid-cuetrials, it appeared either at the uncued end of the cued rectan-gle, or in the uncued rectangle. The distance between thecued location and the location where the target appeared wasidentical in both invalid-cue conditions. On invalid-cue trials,targets were detected faster when they belonged to the sameobject as the cue, rather than to the other object. Severalreplications were reported, with detection (e.g., Lamy & Tsal,2000; Vecera, 1994) as well as identification tasks (e.g.,Lamy & Egeth, 2002; Moore, Yantis, & Vaughan, 1998).

Although some individual studies have been criticizedor proved difficult to replicate and limiting conditions forobject-based selection have been identified (Lamy & Egeth,in press; Watson & Kramer, 1999), the overall picture thatemerges from this selective review is that the segmentation of

Figure 10.2 Sample stimuli and results from Experiment 1 of Watson and Kramer (1999). Each wrench in thetwo upper panels is homogeneously colored, and thus, according to Palmer and Rock (1994), may be character-ized as a single uniformly connected (UC) region. The wrenches in the two lower panels, having stippled handlesbetween solid black ends, each consist of multiple (i.e., three) UC regions. Subjects searched the display for thepresence of two targets: an open end (shown as the upper right end in each panel), and a bent end (shown on theupper left end on the different-wrench examples, and the lower right end of the same-wrench examples). Meanreaction-time differences are shown on the right of the figure. They show a same-object effect for the single-UCwrenches, but not for the wrenches composed of multiple UC regions. Source: Reprinted from Watson andKramer (1999), with permission of the Psychonomic Society.

Stimulus Examples Same Object Benefit

Different - Same WrenchReaction Time (RT)

DifferentWrench

SameWrench

66 msec.

RT (msec.)

�4 msec.

�5 15 35 55 75

UCRegions

Single

Multiple

Efficiency of Selection 273

the visual field into perceptual groups imposes constraints onattentional selection. It is important to note, however, thatthis conclusion does not necessarily imply that groupingprocesses are preattentive. Indeed, in all the studies surveyedabove, at least one part of the relevant object (i.e., of the per-ceptual group for which object-based effects were measured)was attended. As a result, one may conceive of the possibilitythat attending to an object part causes other parts of this ob-ject to be attended.

For this reason, a safer avenue to investigate whethergrouping requires attention may be to measure grouping ef-fects when the relevant perceptual group lies entirely outsidethe focus of attention. The studies pertaining to this issue willbe discussed in the section on “Preattentive and AttentiveProcessing.”

Capture of Attention by Irrelevant Stimuli

Goal-directed or top-down control of attention refers to theability of the observer’s goals or intentions to determinewhich regions, attributes, or objects will be selected forfurther visual processing. Most current models of attentionassume that top-down selectivity is modulated by stimulus-driven (or bottom-up) factors, and that certain stimulus prop-erties are able to attract attention in spite of the observer’seffort to ignore them. Several models, such as the guidedsearch model of Cave and Wolfe (1990), posit that an item’soverall level of attentional priority is the sum of its bottom-upactivation level and its top-down activation level. Bottom-upactivation is a measure of how different an item is from itsneighbors. Top-down activation (Cave & Wolfe, 1990) or

inhibition (Treisman & Sato, 1990) depends on the degree ofmatch between an item and the set of target properties speci-fied by task demands. However, the relative weight allocatedto each factor and the mechanisms responsible for this allo-cation are left largely unspecified. Curiously enough, no par-ticular effort has been made to isolate the effects on visualsearch of bottom-up and top-down factors, which were typi-cally confounded in the experiments held to support thesetheories (see Lamy & Tsal, 1999, for a detailed discussion).For instance, the fact that search for feature singletons is effi-cient has been demonstrated repeatedly (e.g., Egeth, Jonides,& Wall, 1972; Treisman & Gelade, 1980) and has beentermed pop-out search (or parallel feature search). It is oftenassumed that this phenomenon reflects automatic capture ofattention by the feature singleton. However, in typical pop-out search experiments, the singleton target is both task rele-vant and unique. Thus, it is not possible to determine in thesestudies whether efficient search stems from top-down factors,bottom-up factors, or both (see Yantis & Egeth, 1999).

Recently, new paradigms have been designed that allowone to disentangle bottom-up and top-down effects more rig-orously. The general approach has been to determine the ex-tent to which top-down factors may modulate the ability of anirrelevant salient item to capture attention. Discontinuities,such as uniqueness on some dimension (e.g., color, shape,orientation) or abrupt changes in luminance, are typicallyused as the operational definition of bottom-up factors orstimulus salience. Based on the evidence that has accumu-lated in the last decade or so, two opposed theoretical pro-posals have emerged. Some authors have suggested thatpreattentive processing is driven exclusively by bottom-upfactors such as salience, with a role for top-down factors onlylater in processing (e.g., M. S. Kim & Cave, 1999; Theeuwes,Atchley, & Kramer, 2000). Others have proposed that atten-tional allocation is always ultimately contingent on top-down attentional settings (e.g., Bacon & Egeth, 1994; Folk,Remington, & Johnston, 1992). A somewhat intermediateviewpoint is that pure, stimulus-driven capture of attention isproduced only by the abrupt onset of new objects, whereasother salient stimulus properties do not summon attentionwhen they are known to be irrelevant (e.g., Jonides & Yantis,1988). Several sets of findings have shaped the current stateof the literature on how bottom-up and top-down factors af-fect attentional priority.

Beginning in the early 1990s, Theeuwes (e.g., 1991, 1992;Theeuwes et al., 2000) carried out several experiments sug-gesting that attention is captured by the element with thehighest bottom-up salience in the display, regardless ofwhether this element’s salient property is task relevant. Cap-ture was measured as slower performance in parallel search

Figure 10.3 Examples of typical sequences of events in Experiments 1 and2 of the study by Egly, Driver, and Rafal (1994). The white lines in the cuedisplay represent the cue. The filled end of a bar represents the target. Thetarget for the valid trial is in the same spatial location (upper right) as the cue.There are two types of invalid trials. In one, the target is the on the same baras the cue, but at the opposite end, and thus requires a within-object shift ofattention from the preceding cue. In the other, the target is on the uncued bar;this target requires a between-objects switch of attention from the cue. Notethat the distance between the target and the cue is equal in the two types ofinvalid trial.

differentobject

sameobject

or

FIXATION CUE ISI TARGET(invalid)

TARGET(valid)

� � �

�

�

�

274 Attention

even when the target’s unique feature value was known (seePashler, 1988a, for an earlier report of this effect). Theeuwesconcluded that when subjects are engaged in a parallelsearch, perfect top-down selectivity based on stimulus fea-tures (e.g., red or green) or stimulus dimensions (e.g., shapeor color) is not possible.

Bacon and Egeth (1994) questioned this conclusion.Using a distinction initially suggested by Pashler (1988a),they proposed that in Theeuwes’s (1992) experiment, twosearch strategies were available: (a) singleton detection mode,in which attention is directed to the location with the largestlocal feature contrast, and (b) feature search mode, which en-tails directing attention to items possessing the target visualfeature. Indeed, the target was defined as being a singletonand as possessing the target attribute. If subjects used single-ton detection mode, both relevant and irrelevant singletonscould capture attention, depending on which exhibited thegreatest local feature contrast. To test this hypothesis, Baconand Egeth (1994) designed conditions in which singleton de-tection mode was inappropriate for performing the task. Asa result, the disruption caused by the unique distractor dis-appeared. They concluded that irrelevant singletons may ormay not cause distraction during parallel search for a knowntarget, depending on the search strategy employed.

Another set of experiments revealed that abrupt onsets doproduce involuntary attentional capture (Hillstrom & Yantis,1994; Jonides & Yantis, 1988), whereas feature singletons ondimensions such as color and motion do not (e.g., Jonides &Yantis, 1988). These authors concluded that (a) abrupt onsetsare unique in their ability to summon attention to their loca-tion automatically, and (b) feature singletons do not captureattention when they are task irrelevant.

The idea that the ability of a salient stimulus to cap-ture attention depends on top-down settings—specifically,on whether subjects use singleton detection mode or featuresearch mode—is consistent with the contingent attentional cap-ture hypothesis (e.g., Folk et al., 1992).According to this theory,attentional capture is ultimately contingent on whether a salientstimulus property is consistent with top-down attentionalcontrol settings. The settings are assumed to reflect current be-havioral goals determined by the task to be performed. Once theattentional system has been configured with appropriate controlsettings, a stimulus property that matches the settings willproduce “on-line” involuntary capture to its location. Stimulithat do not match the top-down attention settings will not cap-ture attention.

Folk et al. (1992) provided support for this claim using anovel spatial cuing paradigm. In Experiment 3, for instance,subjects saw a cue display followed by a target display (seeFigure 10.5). They were required to decide whether the target

Figure 10.4 Sample stimuli from the studies of Theeuwes (1991, 1992).The subject always searched for a green circle among green diamonds (twoleft panels; form condition), or among red circles (two right panels; colorcondition), either without a distractor (top panels), or with a distractor (bot-tom panels). The line segment within the target element was horizontal orvertical (subjects had to indicate which); the line segments in the other formswere tilted 22.5 deg from horizontal or vertical. Source: Reprinted fromTheeuwes (1992), with permission of the Psychonomic Society.

form color

greenred

when an irrelevant salient object was present. For instance,Theeuwes (1991, 1992) presented subjects with displays con-sisting of varying numbers of colored circles and diamondsarranged on the circumference of an imaginary circle (seeFigure 10.4). A line segment varying in orientation appearedinside each item, and subjects were required to determine theorientation of the line segment within a target item. In onecondition, the target item was defined by its unique form(e.g., it was the single green diamond among green circles).In another condition, it was defined as the color singleton(e.g., it was the single red square among green squares). Onhalf of the trials, an irrelevant distractor unique on an irrele-vant dimension might be present. For instance, when thetarget item was a green diamond among green circles, a redcircle was present. Theeuwes (1991) found that the presenceof the irrelevant singleton slowed reaction times (RTs) signif-icantly. However, this effect occurred only when the irrele-vant singleton was more salient than the singleton target,suggesting that items are selected by order of salience. In alater study, Theeuwes (1992) reported distraction effects

Efficiency of Selection 275

Figure 10.5 Sample cue displays and target displays used to investigatecontingent attentional capture. In these examples, the cues appear in the left-hand location and the targets in the right-hand location (thus any trials com-posed from these particular components would be considered invalid trials).See text for further details. Source: Reprinted from Folk, Remington,and Johnston (1992), with permission of the American PsychologicalAssociation.

Color cue

Onset cue

Color target

Onset target

was an x or an “=” sign. The target was defined either as acolor singleton target (e.g., the single red item among whiteitems) or as an onset target (i.e., a unique abruptly onset itemin the display). Two types of distractors were used. A colordistractor consisted of four colored dots surrounding a poten-tial target location, and arrays of white dots surrounded theremaining three locations. An onset distractor consisted of aunique array of four white dots surrounding one of the poten-tial target locations. The two distractor types were factoriallycombined with the two target types, with each combinationpresented in a separate block. The locations of the distractorand target were uncorrelated. The authors reasoned that if adistractor were to capture attention, a target sharing its loca-tion would be identified more rapidly than a target appearingat a different location. Thus, they measured capture as thedifference in performance between conditions in which dis-tractors appeared at the target location versus nontarget loca-tions. The question was whether capture would depend on thematch between the salient property of the distractor and theproperty defining the target. The results showed that it did:Whereas capture was found when the distractor and targetshared the same property, virtually no capture was observedwhen they were defined by different properties.

The foregoing discussion of attentional capture suggeststhat the conditions under which involuntary capture occurs

remain controversial. Studies that reached incompatible con-clusions usually presented numerous procedural differences.For instance, Folk (e.g., Folk et al., 1992) and Yantis (e.g.,Yantis, 1993) disagree on what status should be assigned tonew (or abruptly onset) objects. Yantis claims that abrupt on-sets capture attention irrespective of the observer’s inten-tions, whereas Folk argues that involuntary capture by abruptonsets happens only when subjects are set to look for onsettargets. Note, however, that Yantis’s experiments typically in-volved a difficult search, for instance, one in which the targetwas a specific letter among distracting letters (e.g., Yantis &Jonides, 1990) or a line differing only slightly in orientationfrom surrounding distractors (e.g., Yantis & Egeth, 1999). Incontrast, Folk’s subjects typically searched for, say, a red tar-get among white distractors—that is, for a target that sharplydiffered from the distractors on a simple dimension (e.g.,Folk et al., 1992). Thus, the two groups of studies differed asto how much top-down guidance was available to find the tar-get. This factor may possibly account for the better selectiv-ity obtained in Folk’s studies. Further research is needed tosettle this issue.

The main point of agreement seems to be that an irrelevantfeature singleton will not capture attention automatically whenthe task does not involve searching for a singleton target. Thisfinding has been obtained using three different paradigms,under which attentional capture was gauged using differentmeasures: a difference between distractor-present versus dis-tractor-absent trials (Bacon & Egeth, 1994); a difference be-tween trials in which the target and cue occupy the same versusdifferent locations in spatial cueing tasks (e.g., Folk et al.,1992); and the difference between trials in which the target andsalient item do versus do not coincide (e.g., Yantis & Egeth,1999). Although most of the evidence provided by Theeuwes(e.g., 1992) for automatic capture was drawn from studies inwhich the target was a singleton, his position on whether cap-ture occurs when the target is not a singleton is not entirelyclear (see, e.g., Theeuwes & Burger, 1998).

Note, however, that in the current state of the literature,the implied distinction between singleton detection mode,in which any salient distractor will capture attention, andfeature search mode, in which only singletons sharing atask-relevant feature will capture attention, suffers from twoproblems.

First, it is based on the yet-untested assumption that thesingleton detection mode of processing is faster or less cogni-tively demanding than is the feature search mode. Indeed, oneobserves that subjects will use the feature search mode only ifthe singleton detection mode is not an option. For instance,when the strategy of searching for the odd one out is notavailable (e.g., Bacon & Egeth, 1994, Experiments 2 & 3),

276 Attention

an irrelevant singleton does not capture attention. However,the same irrelevant singleton does capture attention whensubjects search for a singleton target with a known feature(e.g., Bacon & Egeth, 1994, Experiment 1). Capture by the ir-relevant singleton occurs despite the fact that using the sin-gleton detection mode will tend to guide attention first towarda salient nontarget on 50% of the trials (or even on 100% ofthe trials; see M. S. Kim & Cave, 1999), whereas using thefeature search mode will tend to guide attention directly tothe target on 100% of the trials. The intuitive explanation forthe fact that subjects use a strategy that is nominally less effi-cient is that the singleton-detection processing mode itselfmust be structurally more efficient. Yet, no study to date hasput this assumption to test.

Second, in studies in which subjects must look for aunique target with a known feature, there is often an elementof circularity in inferring from the data which processingmode subjects use. Indeed, if an irrelevant singleton capturesattention, then the conclusion is that subjects used the single-ton detection mode. If, in contrast, no capture is observed,the conclusion is that they used the feature search mode.However, the factors that induce subjects to use one moderather than the other when both modes are available remainunspecified.

Selection by Location and Other Features

The foregoing section was concerned with factors that limitselectivity. Next, we turn to a description of the mechanismsunderlying the different ways by which attention can be di-rected toward to-be-selected or relevant areas or objects.

Selection by Location

“Attention is quite independent of the position and accom-modation of the eyes, and of any known alteration in theseorgans; and free to direct itself by a conscious and voluntaryeffort upon any selected portion of a dark and undifferencedfield of view” (von Helmholtz, 1871, p. 741, quoted byJames, 1890/1950, p. 438). Since this initial observation wasmade, a large body of research has investigated people’s abil-ity to shift the locus of their attention to extra-foveal lociwithout moving their eyes (e.g., Posner, Snyder, & Davidson,1980), a process called covert visual orienting (Posner,1980).

Covert visual orienting may be controlled in one of twoways, one involving peripheral (or exogenous) cues, and theother, central (or endogenous) cues. Peripheral cues tradi-tionally involve abrupt changes in luminance—usually,abrupt object onsets, which on a certain proportion of the

trials appear at or near the location of the to-be-judged target.With central cues, knowledge of the target’s location is pro-vided symbolically, typically in the center of the display (e.g.,an arrow pointing to the target location). Numerous experi-ments have shown that detection and discrimination of a tar-get displayed shortly after the cue is improved more on validtrials—that is, when this target appears at the same locationas the cue (peripheral cues) or at the location specified by thecue (central cues)—than on invalid trials, in which the targetappears at a different location. Some studies also include neu-tral trials or no-cue trials, in which none of the potential tar-get locations is primed (but see Jonides & Mack, 1984, forproblems associated with the choice of neutral cues). Neutraltrials typically yield intermediate levels of performance. Pe-ripheral and central cues have been compared along two mainavenues.

Some studies have focused on differences in the way at-tention is oriented by each type of cue. The results from thisline of research have suggested that peripheral cues captureattention automatically (but see the earlier section, “Captureof Attention by Irrelevant Stimuli,” for a discussion of thisissue), whereas attentional orienting following a central cueis voluntary (e.g., Müller & Rabbitt, 1989; Nakayama &Mackeben, 1989). Moreover, attentional orienting to the cuedlocation was found to be faster with peripheral cues than withcentral cues. For instance, in Muller and Rabbitt’s (1989)study, subjects had to find a target (T ) among distractors (+)in one of four boxes located around fixation. The central cuewas an arrow at fixation, pointing to one of the four boxes.The peripheral cue was a brief increase in the bright-ness of one of the boxes. With peripheral cues, costs andbenefits grew rapidly and reached their peak magnitudes atcue-to-target onset asynchronies (SOAs) in the range of 100–150 ms. With central cues, maximum costs and benefits wereobtained for SOAs of 200–400 ms.

Other studies have focused on differences in informationprocessing that occur as a consequence of the allocation of at-tention by peripheral versus central cues. Two broad classesof mechanisms have been proposed to describe the effects ofspatial cues. According to the signal enhancement hypothesis(e.g., Henderson, 1996), attention strengthens the stimulusrepresentation by allocating the limited capacity available forperceptual processing. In other words, attention facilitatesperceptual processing at the cued location. According to theuncertainty or noise reduction hypothesis (e.g., Palmer,Ames, & Lindsay, 1993) spatial cues allow one to excludedistractors from processing by monitoring only the relevantlocation rather than all possible ones. Thus, cueing attentionto a specific location reduces statistical uncertainty or noiseeffects, which stem from information loss and decision

Efficiency of Selection 277

limits, not from changes in perceptual sensitivity or limits ofinformation-processing capacity.

In order to test the two hypotheses against each other, sev-eral investigators have sought to determine whether spatialcueing effects would be observed when the target appears in anotherwise empty field. The signal enhancement hypothesispredicts such effects, as the allocation of attentional resourcesat the cued location should facilitate perceptual processing atthat location, even in the absence of noise. In contrast, thenoise reduction hypothesis predicts no cueing effects with sin-gle-element displays, because no spatial uncertainty or noisereduction should be required in the absence of distractors.

This line of research has generated conflicting findings,with reports of small effects (Posner, 1980), significanteffects (e.g., Henderson, 1991) or no effect (e.g., Shiu &Pashler, 1994). Relatively subtle methodological differenceshave turned out to play a crucial role. For instance, Shiuand Pashler (1994) criticized earlier single-target studies(Henderson, 1991) on the grounds that the masks presented ateach potential location after the target display may have beenconfusable with the target, thus making the precue useful inreducing the noise associated with the masks. They compareda condition in which masks were presented at all potential lo-cations vs. a condition with a single mask at the target loca-tion. Precue effects were found only in the former condition,supporting the idea that these reflect noise reduction ratherthan perceptual enhancement. However, recent evidenceshowed that spatial cueing effects can be found with a singletarget and mask, and are larger with additional distractors ormasks. These findings suggest that attentional allocation byspatial precues leads both to signal enhancement at the cuedlocation and noise reduction (e.g., Cheal & Gregory, 1997;Henderson, 1996).

Most of the reviewed studies employed informative pe-ripheral cues, which precludes the possibility of determiningwhether the observed effects of attentional facilitationshould be attributed to the exogenous or to the endogenouscomponent of attentional allocation, or to both. Studiesthat employed non-informative peripheral cues (Henderson,1996; Luck & Thomas, 1999) showed that these lead to bothperceptual enhancement and noise reduction. Recently, Luand Dosher (2000) directly compared the effects of periph-eral and central cues and reported results suggesting a noisereduction mechanism of central precueing and a combinationof noise reduction and signal enhancement for peripheralcueing.

To conclude, the current literature points to notable differ-ences in the way attention is oriented by peripheral vs. centralcues, as well as differences in information processing whenattention is directed by one type of spatial cue vs. the other.

Is Location Special?

The idea that location may deserve a special status in thestudy of attention has generated a considerable amount of re-search, and the origins of this debate can be traced back to thenotion that attention operates as a spotlight (e.g., Broadbent,1982; Eriksen & Hoffman, 1973; Posner et al., 1980), whichhas had a major influence on attention research. According tothis model, attention can be directed only to a small contigu-ous region of the visual field. Stimuli that fall within thatregion are extensively processed, whereas stimuli locatedoutside that region are ignored. Thus, the spotlight model—as well as models based on similar metaphors, such aszoom lenses (e.g., Eriksen & Yeh, 1985) and gradients (e.g.,Downing & Pinker, 1985; LaBerge & Brown, 1989)—endows location (or space) with a central role in the selectionprocess. Later theories making assumptions that markedlydepart from spotlight theories also assume an important rolefor location in visual attention (see Schneider, 1993 for areview). These include for instance Feature IntegrationTheory (Treisman & Gelade, 1980), the Guided Search model(Cave & Wolfe, 1990; Wolfe, 1994), van der Heidjen’s model(1992, 1993), and the FeatureGate model (Cave, 1999).

A comprehensive survey of the debate on whether or notlocation is special is beyond the scope of the present en-deavor (see for instance, Cave & Bichot, 1999; Lamy & Tsal,2001, for reviews of this issue). Here, two aspects of thisdebate will be touched on, which pertain to the efficiency ofselection. First, we shall briefly review the studies in whichselectivity using spatial vs. non-spatial cues is compared.Then, the idea that selection is always ultimately mediated byspace, which entails that selection by location is intrinsicallymore direct, will be contrasted with the notion that attentionselects space-invariant object-based representations.

Selection by Features Other Than Location. Numer-ous studies have shown that advance knowledge about anon-spatial property of an upcoming target can improve per-formance (e.g., Carter, 1982). Results arguing against theidea that attention can be guided by properties other than lo-cation are typically open to alternative explanations (seeLamy & Tsal, 2001, for a review). For instance, Theeuwes(1989) presented subjects with two shapes that appeared si-multaneously on each side of fixation. The target was definedas the shape containing a line segment, whereas the distractorwas the empty shape. Subjects responded to the line’s orien-tation. The target was cued by the form of the shape withinwhich it appeared, or by its location. Validity effects wereobtained with the location cue but not with the form cue.The author concluded that advance knowledge of form

278 Attention

cannot guide attention. Note however, that it may havebeen easier for subjects to look for the filled shape, that is,to use the “defining attribute” (Duncan, 1985), rather thanto use the form cue. In this case, subjects may simply nothave used the cue, which would explain why it had no effect.According to this logic, using the location cue was easierthan looking for the filled shape, but looking for the filledshape was easier than using the form cue. Thus, whereasTheeuwes’s finding indicates that location cueing may bemore efficient than form cueing, it does not preclude the pos-sibility that form cues may effectively guide attention whenno other, more efficient strategy is available.

Whereas it is generally agreed that spatial cueing is moreefficient than cueing by other properties, there has been somedebate as to whether qualitative differences exist between at-tentional allocation using one type of cue vs. the other (e.g.,Duncan, 1981; Tsal, 1983). It seems that non-spatial cues dif-fer from peripheral spatial cues in that they only prioritize theelements possessing the cued property rather than improvingtheir perceptual representation. Moore and Egeth (1998) re-cently presented evidence showing that “feature-based atten-tion failed to aid performance under ‘data-limited’ conditions(i.e., those under which performance was primarily affectedby the sensory quality of the stimulus), but did affect perfor-mance under conditions that were not data-limited.” More-over, in several physiological studies that compared theevent-related potentials (ERP) elicited by stimuli attended onthe basis of location vs. other features, a qualitatively differ-ent pattern of activity was found for the two types of cues,which was taken to indicate that selection by location mayoccur at an earlier stage than selection by other properties(e.g., Hillyard & Munte, 1984; Näätänen, 1986).

Is Selection Mediated by Space? The idea that selec-tion is always ultimately mediated by space, as is assumed innumerous theories of attention, has been challenged by re-search showing that attention is paid to space-invariantobject-based representations rather than to spatial locations.Studies favoring the space-based view typically manipulatedonly spatial factors. The reasoning was that if spatial effectscan be found when space is task irrelevant, then selectionmust be mediated by space, and does not therefore operate onspace-invariant representations. In contrast, in studies sup-porting the space-invariant view, spatial factors were usuallykept constant and objects were separated from their spatiallocation via motion. In spite of intensive investigation, noconsensus has yet emerged.

It is important to make it clear that the body of researchconcerned with the effects of Gestalt grouping on the distrib-ution of attention that was reviewed earlier is not relevant

here. Both issues are generally conflated under the generalterm of “object-based selection.” However, whether attentionselects spatial or spatially-invariant representations concernsthe medium of selection, whereas effects of grouping on at-tention speak to the efficiency of selection (see Lamy & Tsal,2001; Vecera, 1994; Vecera & Farah, 1994, for further expli-cation of this distinction).

One of the most straightforward methods used to investi-gate whether selection is fundamentally spatial is to havesubjects attend to an object that happens to occupy a certainlocation in a first display and then attend to a different objectoccupying either the same or a different location in a subse-quent display. With this procedure, sometimes referred to asthe “post-display probe technique” (e.g., Kramer, Weber, &Watson, 1997), an advantage in the same-location conditionis taken to support the idea that selection is space-based. Thecrux of this method is that it shows spatial effects in taskswhere space is utterly irrelevant to the task at hand. For in-stance, Tsal and Lavie (1993, Experiment 4) showed thatwhen subjects had to attend to the color of a dot (its locationbeing task irrelevant), they responded faster to a subsequentprobe when it appeared in the location previously occupiedby the attended dot than in the alternative location (see M. S.Kim & Cave, 1995, for similar results).

Following a related rationale, other authors used rapidserial visual presentation (RSVP) tasks (e.g., McLean,Broadbent, & Broadbent, 1983) or partial report tasks (e.g.,Butler, Mewhort, & Tramer, 1987) and showed that whensubjects have to report an item with a specific color, near-location errors are the most frequent. In the same vein, Tsaland Lavie (1988) showed that when required to report oneletter of a specified color and then any other letters they couldremember from a visual display, subjects tended to report let-ters adjacent to the first-reported letter more often than lettersof the same (relevant) color (see van der Heijden, Kurvink,de Lange, de Leeuw, & van der Geest, 1996, for a criticismand Tsal & Lamy, 2000, for a response). These results suggestthat selecting an object by any of its properties is mediated bya spatial representation.

Other investigators attempted to demonstrate that selec-tion is mediated by space by showing effects of distance onattention. In early studies, interference was found to be re-duced as the distance between target and distractors increased(e.g., Gatti & Egeth, 1978). Attending to two stimuli was alsofound to be easier when these were close together rather thandistant from each other (e.g., Hoffman & Nelson, 1981).More recent studies showed that distance modulates same-vs.-different object effects, as the difficulty in attending totwo objects increases with the distance between these objects(e.g., Kramer & Jacobson, 1991; Vecera, 1994. See Vecera &

Efficiency of Selection 279

Farah, 1994, for a failure to find distance effects on objectselection, and Kramer et al., 1997; Vecera, 1997, for a dis-cussion of these results). There is some contrary evidence,suggesting that performance gets better as the separation be-tween attended elements increases (e.g., Bahcall & Kowler,1999; Becker, 2001) and still other findings showing that per-formance is unaffected by the separation between attendedstimuli (e.g., Kwak, Dagenbach, & Egeth, 1991).

The experimental strategy of manipulating distance todemonstrate that selection is mediated by space has been crit-icized on several grounds. For instance, distance effects in di-vided attention tasks may only reflect the effects of groupingby proximity. That is, when brought closer together, two ob-jects may be perceived as a higher-order object (e.g., Duncan,1984). Accordingly, distance effects are attributed to effectsof grouping on the distribution of attention and say nothingabout whether or not the medium of attention is spatial. Intasks involving a shift of attention over small vs. large dis-tances, the assumption underlying the use of a distancemanipulation is that attention moves in an analog fashionthrough visual space, the time needed for attention to movefrom one location to another being proportional to the dis-tance between them. However, this assumption may be un-warranted (e.g., Sperling & Weichselgartner, 1995).

Support for the Space-Invariant View. Whether at-tention may select from space-invariant object-based repre-sentations has been investigated by separating objects fromtheir locations via motion. Kahneman, Treisman, and Gibbs(1992) found that the focusing of attention on an object se-lectively activates the recent history of that object (i.e., itsprevious states) and facilitates recognition when the currentand previous states of the object match. They found thismatching process, called “reviewing,” to be successful onlywhen the objects in the preview and probing displays sharedthe same “object-file,” namely, when one object was perceivedto move smoothly from one display to the other. This findingis typically taken to show that attention selects object-files,that is, representations that maintain their continuity in spiteof location changes (e.g., Kanwisher & Driver, 1992).

Further support for the idea that attention operates inobject-based coordinates comes from experiments by Tipperand his colleagues. They used the inhibition of return para-digm (e.g., Tipper, Weaver, Jerreat, & Burak, 1994) and thenegative priming paradigm (Tipper, Brehaut, & Driver,1990), as well as measurements of the performance of neglectpatients (Behrmann & Tipper, 1994). Inhibition of returnstudies show that it is more difficult to return one’s attentionto a previously attended location. (Immediately after a spatiallocation is cued, a stimulus is relatively easy to detect at the

cued location. However, after a cue-target SOA of about300 ms, target detection is relatively difficult at the cued loca-tion. This is known as inhibition of return.) Negative primingexperiments demonstrate that people are slower to respond toan item if they have just ignored it. (For a further discussionof negative priming, see the chapters by Proctor & Vu andMcNamara & Holbrook in this volume.) Finally, the neurobi-ological disorder called unilateral neglect is characterized bythe patients’ failure to respond or orient to stimuli on the sidecontralateral to a lesion. Although early studies suggested thatall three phenomena are associated with spatial locations(e.g., Posner & Cohen, 1984; Tipper, 1985; and Farah, Brunn,Wong, Wallace, & Carpenter, 1990, respectively), recentstudies using moving displays showed that the attentional ef-fects revealed by each of these experimental methods can beassociated with object-centered representations.

Lamy and Tsal (2000, Experiment 3) used a variant of Eglyet al.’s (1994) task. Subjects had to detect a target at one of thefour ends of two objects, differing in color and shape. A pre-cue appeared at one of the four ends and indicated the locationwhere the target was most likely to show up. To dissociate thecued object from its location, the two objects were made toexchange locations between the cueing and target displays, bymoving smoothly, on half of the trials. Reaction times werefaster at the uncued location within the cued object than at anequally distant location within the uncued object, thus indi-cating that attention followed the cued object-file.

Conclusions. To summarize, in studies that measuredonly space-based effects using either the distance manipula-tion or the post-display probe technique, it was typicallyfound that selection is mediated by space. In studies that mea-sured the cost of redirecting attention to the same vs. a differ-ent object-file using moving objects while keeping spatialfactors constant, attention was typically found to follow theobject initially attended as it moved. Note that the strongestsupport for the view that selection is mediated by spacecomes from studies in which response to a new object wasfound to be faster if this object occupied the location of a pre-viously attended object even when space was irrelevant to thetask. Thus, in these studies, the object initially attended wasno longer present in the subsequent display, where attentionaleffects were measured: A different object typically replacedit. Such findings may therefore only indicate that space-basedselection prevails when the task is such that object continuityis systematically disrupted. In other words, selection may bespace-based only under this specific condition, which doesnot abound in a natural environment.

On the other hand, support for the idea that selection oper-ates on space-invariant representations of objects comes from

280 Attention

studies showing that attending to an object entails that atten-tional effects remain associated with this object as it moves.However, space and object-file effects may not be as antithet-ical as is usually assumed. Finding that attention follows thecued object-file as it moves does not necessarily argue againstthe idea that selection is mediated by space. Attention maysimply accrue to the locations successively occupied by themoving object (e.g., Becker & Egeth, 2000). As yet, no em-pirical data have been reported that preclude this possibility.

PREATTENTIVE AND ATTENTIVE PROCESSING

As was mentioned in the introductory part of this chapter,inquiring which processes are not contingent on capacitylimitations for their execution amounts to inquiring whichprocesses are preattentive, that is, do not require attention.“What does the preattentive world look like? We will neverknow directly, as it does not seem that we can inquire aboutour perception of a thing without attending to that thing”(Wolfe, 1998, p. 42). Therefore, it takes ingenious experi-mental designs to investigate the extent to which unattendedportions of the visual field are processed.

Two general empirical strategies have traditionally beenused to address this question, and differ somewhat in the un-derlying definition of “preattentiveness” they adopt. In someparadigms (e.g., visual search), whatever processes do not re-quire focused attention and can be performed in parallel withattention widely distributed over the visual field are consid-ered to be preattentive. In other paradigms (e.g., dual task),preattentive processes are those processes that can proceedwithout attention, that is, when attentional resources are ex-hausted by some other task. As we shall see, interpretingresults obtained pertaining to preattentive processing hasproved to be tricky.

Distributed Attention Paradigms

Visual Search

In a standard visual search experiment, the subject might beasked to indicate whether a specified target is present or ab-sent, or which of two possible targets is present among anarray of distractors. The total number of items in the display,known as the set size or display size, usually varies from trialto trial. The target is typically present on 50% of the trials, thedisplay containing only distractors on the remaining trials.On each trial, subjects have to judge whether a target is pre-sent. In studies measuring reaction times, the search displayremains visible until subjects respond. Of chief interest is the

way reaction times vary as a function of set size on target-present and target-absent trials. In studies measuring accu-racy, search displays are presented briefly and then masked.Accuracy can be plotted as a function of set size to reveal theprocesses underlying search. A common alternative approachis to determine the exposure duration (typically, the asyn-chrony between the onsets of the search display and of a sub-sequent masking display) required to achieve some fixedlevel of accuracy (e.g., 75% correct).

If finding the target (i.e., distinguishing it from the distrac-tors) involves processes that do not require attention and areperformed in parallel over the whole display, one expects toobserve parallel search. With studies measuring reactiontime, this means that the number of distractors present in thedisplay should not affect performance; with studies measuringaccuracy, this means that beyond a relatively short SOA, in-creasing the time available to inspect the display should notimprove performance. Thus, parallel search is held to bediagnostic of preattentive (i.e., parallel, resource-free) pro-cessing. If, in contrast, distinguishing the target from the dis-tractors involves processes that do require attention, thenattention must be directed to the items one at a time (or perhapsto one subset of them at a time), until the target is found. In thiscase, the time required to find the target increases as the num-ber of distractors increases. Moreover, if search is terminatedas soon as the target is found, the target should be found, on av-erage, halfway through the search process. Thus, search slopesfor target-absent trials should be twice as large as for target-present trials (Sternberg, 1969). In studies measuring accu-racy, if search requires attention, the more items in the displaythe longer the exposure time necessary to find the target.

This rationale was criticized very early on (Luce, 1986;Townsend, 1971; Townsend & Ashby, 1983). On the onehand, slopes are usually shallow, perhaps 10 ms per item,rather than null. In principle, they could reflect the operationof a serial mechanism that processes 100 items every second.However, such fast scanning is held to be physiologically notfeasible (Crick, 1984).

On the other hand, linear search functions do not neces-sarily reflect serial processing. They are consistent withcapacity-limited parallel processing, in which all items areprocessed at once, although the rate at which information ac-cumulates at each location for the presence of the target or ofa nontarget item decreases as the number of additional com-parisons concurrently performed increases (Murdock, 1971;Townsend & Ashby, 1983).

Linear search functions are also compatible with unlimited-capacity parallel processing, in which set size affects the dis-criminability of elements in the array rather than processingspeed per se. According to this view, the risk of confusing the

Preattentive and Attentive Processing 281

target with a distractor increases as the number of elements in-creases, due to decision processes or to sensory processes(Palmer et al., 1993).

Finally, it has been argued that even the steepest serialslopes cannot reflect serial item-by-item attentional scanning.Whereas these range from 40 to 100 ms per item, Duncan,Ward, and Shapiro (1994) have claimed that attention mustremain focused on an object for several hundred millisecondsbefore being shifted to another object. They referred to thisperiod as the attentional dwell time. However, Moore, Egeth,Berglan, and Luck (1996) have shown that the long estimatesof dwell time were caused, at least in part, by the use ofmasked targets.

Simultaneous versus Successive Presentation

Considering the complexities involved in interpreting searchslopes, several investigators have explored the ability of indi-viduals to discriminate between two targets in displays of afixed size in which the critical manipulation involves the waythe stimuli are presented over time. These experiments com-pare a condition in which all of the stimuli are presentedsimultaneously with a condition in which they are presentedsequentially. (They may be presented one at a time or inlarger groups.) Each stimulus is followed by a mask. Thelogic is that if capacity is limited, then it should be more dif-ficult to detect a target when all of the stimuli are presented atthe same time than when they are presented in smallergroups, which would permit more attention to be devoted toeach item.

Shiffrin and Gardner (1972) showed that when a fairlysimple discrimination was involved, such as indicatingwhether a T or an F target was present in a display (thenontargets here were hybrid T-F characters), and the numberof display elements was small (four), then there was good ev-idence of parallel processing with unlimited capacity (seealso Duncan, 1980). However, when the number of elementsin the display was increased (e.g., Fisher, 1984) or the com-plexity of the stimuli was increased, advantages for succes-sive presentation have been observed (e.g., Duncan, 1987;see also Kleiss & Lane, 1986).

Change Blindness

In an interesting variant of a search task, subjects are pre-sented with a display that is replaced with a second displayafter a delay filled with a blank field, and have to indicatewhat, if anything, is different about the second display. Thedisplays can be of any sort, from random displays of dots(Pollack, 1972) to real-life visual events (e.g., Simons &

Levin, 1998). These conditions lead to a wide deployment ofattention over the visual field. The striking result is that sub-jects show very poor performance in detecting the change, aneffect that has been dubbed change blindness.

The change blindness effect is reminiscent of subjects’failure to detect changes that occur during a saccadic eye move-ment (e.g., Bridgeman, Hendry, & Stark, 1975). However, sub-sequent research has shown that it may occur independently ofsaccade-specific mechanisms (Rensink, O’Regan, & Clark,1997). The two paradigms that are most frequently used to in-vestigate the change blindness phenomenon are the flicker par-adigm (Rensink et al., 1997) and the forced-choice detectionparadigm (e.g., Pashler, 1988b; Phillips, 1974).

In the forced-choice detection paradigm, each trial con-sists of one presentation each of an original and a modifiedimage. Only some of the trials contain changes, which makesit possible to use signal detection analyses in addition to mea-suring response latency and accuracy. For instance, Phillips(1974) presented matrices that contained abstract patterns ofblack and white squares and asked subjects to detect changesbetween the first and second displays. When the interstimulusinterval was short (tens of milliseconds) the task was easy be-cause subjects saw either flicker or motion at the locationwhere a change was made. However, when the interstimulusinterval was longer the task became very difficult becauseoffset and onset transients occurred over the entire visualfield and thus could not be used to localize the matrix loca-tions that had been changed.

In the flicker paradigm, the original and the modifiedimage are presented in rapid alternation with a blank screenbetween them. Subjects respond as soon as they detect themodification. The results typically show that subjects almostnever detect changes during the first cycle of alternation, andit may take up to 1 min of alternation before some changes aredetected (Rensink et al., 1997), even though the changes areusually substantial in size (typically about 20 deg.2) and oncepointed out or detected are extremely obvious to theobservers. Moreover, changes to objects in the center of in-terest of a scene are detected more readily than peripheral, ormarginal-interest, changes (Rensink et al.). Rensink et al.concluded that “visual perception of change in an object oc-curs only when that object is given focused attention; in theabsence of such attention, the contents of visual memory aresimply overwritten (i.e., replaced) by subsequent stimuli, andso cannot be used to make comparisons” (p. 372). Based onthe change blindness finding and the results from studies ofvisual integration (e.g., Di Lollo, 1980), Rensink (2000)speculated that the preattentive representation of a scene“formed at any fixation can be highly detailed, but will havelittle coherence, constantly regenerating as long as the light

282 Attention

continues to enter the eyes, and being created anew after eacheye movement” (p. 22).

However, as was noted by Simons (2000), there are otherpossible accounts for the change blindness effect. “For exam-ple, we might retain all of the visual details across views, butnever compare the initial representation to the current per-cept. Or, we might simply lack conscious access to the visualrepresentation (or to the change itself) thereby precludingconscious report of the change” (p. 7). Thus, the finding ofchange blindness does not necessarily imply that the repre-sentation of the initial scene is absent. Further research usingimplicit measures to evaluate the extent to which this repre-sentation is preserved will be useful in order to expand ourknowledge not only concerning the change blindness phe-nomenon but more generally, concerning preattentive visionand the role of attention.

Inattention Paradigms: Dual-Task Experiments

In dual-task experiments designed to explore what processesare preattentive, subjects have to execute a primary task and asecondary task. In some cases (e.g., Mack, Tang, Tuma, Kahn,& Rock, 1992; Rock, Linnett, Grant, & Mack, 1992), the pri-mary task is assumed to exhaust subjects’ processing capaci-ties or to ensure optimal focusing of attention. If subjects cansuccessfully perform the secondary task, then it is concludedthat the processes involved in that task do not require attentionand are therefore preattentive. The studies using this logicusually suffered from memory confounds, as subjects weretypically requested to overtly report what they had seen in thesecondary task displays after performing the primary task.

In other cases (e.g., Joseph, Chun, & Nakayama, 1997;Braun & Sagi, 1990, 1991), performance is compared be-tween a condition in which subjects have to perform both theprimary and the secondary task (a dual-task condition) and acondition in which subjects are required to perform only thesecondary task (a single-task condition). Sometimes an addi-tional single-task control condition is used, in which subjectsare required to perform only the primary task. When agiven task is performed equally well in the single- and dual-task conditions, this performance is taken to indicate thatprocesses involved in the secondary task are preattentive,whereas poorer performance in the dual-task condition isheld to show that these processes require attention. A caveatthat is sometimes associated with this rationale is that theperformance impairment produced by the addition of theprimary task may reflect the cost of making two responsesversus only one, rather than the inability to process the sec-ondary task preattentively. (The results of the studies citedabove are discussed later in this chapter.)

We now proceed to present a few examples of efforts todistinguish between processes that require attention andprocesses that are preattentive.

Further Explorations of Preattentive Processing

Grouping

Is perceptual grouping accomplished preattentively? This hasproven difficult to answer, in part because grouping itself is acomplex concept. For example, Trick and Enns (1997), fol-lowing Koffka (1935, pp. 125–127), distinguish between ele-ment clustering and shape formation. Their research suggeststhat the former is preattentive, whereas the latter requires at-tention. Consider the stimuli in Figure 10.6. In two panelsthe stimuli consist of small diamond shapes made up of con-tinuous lines, while in the other two panels the diamondsare made up of four small dots. Subjects had to determine thenumber of diamonds present in a display; reaction timewas the dependent variable of chief interest. The two panelson the left yielded essentially identical results. The fact thatclusters of dots can be counted as quickly as continuous lineforms, even for small numbers of elements in the subitizingrange (1–3 or 4 items), is consistent with the idea that the dotscomposing the diamonds were clustered preattentively. Forrelated results, see Bravo and Blake (1990). Interestingly,when shape discrimination was required (counting the dia-monds in the face of square distractors, as shown on the rightside of Figure 10.6), the continuous line forms were countedmore efficiently than the stimuli made of dots. This suggeststhat the shape formation process may not be preattentive.

That the shape formation component of grouping may re-quire attention is consistent with a number of experimentsthat suggest grouping outside the focus of attention is not per-ceived (e.g., Ben-Av, Sagi, & Braun, 1992; Mack et al., 1992;Rock et al., 1992), suggesting that attention selects unparsedareas of the visual field and that grouping requires attention.Ben-Av et al. showed that subjects’ performance in discrimi-nating between horizontal and vertical grouping, or in simplydetecting the presence or absence of grouping in the displaybackground, was severely impaired when attention was en-gaged in a concurrent task of form identification of a targetsituated in the center of the screen. Mack et al. obtained sim-ilar results with grouping by proximity and similarity oflightness.

However, the dependent measure in these studies was sub-jects’ conscious report of grouping. The fact that groupingcannot be overtly reported when attention is engaged in a de-manding concurrent task does not necessarily imply thatgrouping requires attention. For instance, failure to report

Preattentive and Attentive Processing 283

No Distractors (Experiment 1) Eight Distractors (Experiment 2)

Lin

e Fo

rms

Dot

For

ms

Figure 10.6 Sample stimuli of the kind used by Trick and Enns (1997) to investigate grouping.

grouping may result from memory failure. That is, groupingprocesses may occur preattentively, with grouping being per-ceived yet not remembered. In order to test this possibility,Moore and Egeth (1997) conducted a study with displays con-sisting of a matrix of uniformly scattered white dots on a graybackground, in the center of which were two black horizontallines (see Figure 10.7). Some of the dots were black, and oncritical trials they were grouped and formed either the Ponzoillusion (Experiments 1 & 2) or the Müller-Lyer illusion (Ex-periment 3). Subjects attended to the two horizontal lines andreported which one was longer. Responses were clearly influ-enced by the two illusions. Therefore, the fact that elementslying entirely outside the focus of attention formed a groupdid affect behavior, indicating that grouping does not requireattention. In a subsequent recognition test, subjects were un-able to recognize the illusion patterns. This result confirmedthe authors’ hypothesis that implicit measures may reveal thatsubjects perceive grouping, whereas explicit measures maynot. (For a further discussion of the consequences of inatten-tion, see the chapter by Banks in this volume.)

Visual Processing of Simple Features versusConjunctions of Features

Treisman’s feature integration theory (FIT; e.g., Treisman &Gelade, 1980; Treisman & Schmidt, 1982) has inspired much

of the research on visual search ever since its inception in theearly 1980s. According to the theory, input from a visualdisplay is processed in two successive stages. During thepreattentive stage, a set of spatiotopically organized mapsis extracted in parallel across the visual field, with eachmap coding the presence of a particular elementary stimulusattribute or feature (e.g., red or vertical). In the second stage,attention becomes spatially focused and serves to glue fea-tures occupying the same location into unified objects.

The phenomenon of illusory conjunctions (e.g., Treisman& Schmidt, 1982; Prinzmetal, Presti, & Posner, 1986;Briand & Klein, 1987) provides empirical support for the FIT.In the experiments of Treisman and Schmidt displays con-sisted of several shapes with different colors flanked by twoblack digits. The primary task was to report the digits, and thesecondary task was then to report the colored shapes. Subjectstended to conjoin the different colors and forms erroneously.For instance, they might report seeing a red square and a bluecircle when in fact a red circle and a blue square had been pre-sent. This finding is thus consistent with the idea that featuresare “free-floating” at the preattentive stage and that focusedattention is needed to correctly conjoin them. However, thefact that subjects were unable to remember how the forms andcolors were combined does not necessarily entail that suchunified representations were not extracted in the absence ofattention. Indeed, an alternative explanation is that illusory

284 Attention

Figure 10.7 Sample stimuli used Moore and Egeth (1997). Subjects had tojudge whether the upper or lower solid black line was longer. On the top is atypical noncritical trial, on the bottom a typical critical trial in which theblack dots are arranged in such a manner as to induce the Ponzo illusion.Source: Reprinted from Moore and Egeth (1997), with permission of theAmerican Psychological Association.

conjunctions may not reflect separate coding of different fea-tures at the preattentive stage but rather the tendency ofmemorial representations of unified objects to quickly disin-tegrate, and more so when no attention is available to main-tain these representations in memory (Virzi & Egeth, 1984;Tsal, 1989). More recent studies suggest that rather than de-riving from imperfect binding of correctly perceived features,illusory conjunctions may stem from target-nontarget confu-sions (Donk, 1999), uncertainty about the location of visualfeatures (Ashby, Prinzmetal, Ivry, & Maddox, 1996) or post-perceptual factors (Navon & Ehrlich, 1995).

A central source of support for FIT also resides in the find-ing that searching for a target that is unique in some elemen-tary feature (e.g., searching for a red target among green andblue distractors) yields fast reaction times and low error ratesthat are largely unaffected by set size (e.g., Treisman &Gelade, 1980; see also Egeth et al., 1972). According to thetheory, features can be detected by monitoring in parallel the

net activity in the relevant feature map (e.g., red). In contrast,searching for a target that is unique only in its conjunction offeatures (e.g., searching for a red vertical line among greenvertical and red tilted lines) yields slower RTs and highererror rates that increase linearly with set size. Attention needsto be focused serially on each item in order to integrate infor-mation across feature modules, because correct feature con-junction is necessary in order to distinguish the target fromthe distractors. This interpretation of the results has been crit-icized on numerous grounds.

As we mentioned earlier, several alternative models showthat parallel and serial processing cannot be directly inferredfrom flat and linear slopes, respectively. Moreover, new find-ings have seriously challenged the parallel versus serial pro-cessing dichotomy originally advocated by FIT. For instance,a number of studies have shown that feature search is not al-ways parallel or effortless. Indeed, feature search was foundto yield steep slopes when distractors were similar to the tar-get or dissimilar to each other (e.g., Duncan & Humphreys,1989; Nagy & Sanchez, 1990). Joseph et al. (1997) furthershowed that even a simple feature search (detecting an orien-tation singleton) that produces flat slopes when executed onits own may be impaired by the addition of a primary taskwith high attentional demands; the data for this experimentappear in the right panel of Figure 10.8. Although Braun(1998; see also Braun & Sagi, 1990) did not replicate Josephet al.’s (1997) results when subjects were well practicedrather than naive, the Joseph et al. findings nevertheless,“seem to rule out a conceivable architecture for the visualsystem in which all feature differences are processed along apathway that has a direct route to awareness, without havingto pass through an attentional bottleneck” (Joseph et al.,1997, p. 807). Thus, Joseph et al.’s results do not challengethe idea that certain feature differences may be extractedpreattentively and only overrule the notion that these differ-ences may be reported without attention.

Other studies demonstrated that some conjunctionsearches are parallel (e.g., Duncan & Humphreys, 1989;Egeth, Virzi, & Garbart, 1984; Wolfe, Cave, & Franzel,1989). For instance, Egeth et al. (1984) showed that subjectswere able to limit their searches to items of a specific color orspecific form. Wolfe et al. (1989) reported shallow searchslopes for targets defined by conjunctions of color and form.Duncan and Humphreys (1989) showed that when target-distractor and distractor-distractor similarity are equated be-tween feature and conjunction search tasks, performance onthese tasks behaves no differently, and concluded that there isnothing intrinsically different between feature and conjunc-tion search (see Duncan & Humphreys, 1992, and Treisman,1992, for a discussion of this idea). Recently, McElree and

Preattentive and Attentive Processing 285

Figure 10.8 These figures show two different forms of the attentional blink. Left panel: Percentage of correct identifications of thesecond target as a function of the temporal lag from the onset of the first target to the onset of the second. Performance at the shortest lagexhibits what has been called lag-1 sparing. Reprinted from Chun and Potter (1995), with permission of the American Psychological As-sociation. Right panel (filled symbols): Same as for left panel, except that performance at the shortest lag does not exhibit lag-1 sparing.The open symbols represent control condition performance on the “second” target when it was the only target in the display. Reprintedfrom Joseph, Chun, and Nakayama (1997), with permission of Nature. Visser, Bischof, and DiLollo (1999) discuss in detail the condi-tions that determine whether the attentional blink will be nonmonotonic or monotonic in form.

A B

0

50

60

70

80

90

100

100 200 300 400 500 600 700

Lag (ms)0 100 200 300 400 500 600 700 800 900 1000

Perc

enta

ge c

orre

ct

20

30

40

50

60

70

80

90

100Pe

rcen

t rep

ort o

f T

2 gi

ven

T1

Dual-task

Single-task

Stimulus Onset Asynchrony

Carrasco (1999) used the response-signal speed-accuracytrade-off (SAT) procedure in order to distinguish between theeffects of set size on discriminability and processing speed inboth feature and conjunction search, and concluded that bothfeature and conjunctions are detected in parallel.

As a result of this spate of inconsistent findings, FIT has un-dergone several major modifications and alternative modelshave been developed, the most influential of which are theguided search model proposed by Cave and Wolfe (1990)and periodically revised by Wolfe (e.g., 1994, 1996) andDuncan and Humphreys’(1989) engagement theory, some-times known as similarity theory. (The guided search modelwas described briefly in the section on “Capture of Attentionby Irrelevant Stimuli.”)

Stimulus Identification

When a subject searches through a display for a target, thenontarget items obviously must be processed deeply enoughto allow their rejection, but this does not necessarily meanthat they are fully identified. For example, in search for adigit among letters, one does not necessarily have to knowthat a character is a G to know that it is not a digit. Thus, it ispossible that some evidence for parallel processing (e.g.,Egeth et al., 1972) may not indicate the ability to identify sev-eral characters in parallel.

Pashler and Badgio (1985) designed a search task notsubject to this shortcoming; they showed several digits

simultaneously and asked subjects to name the highest digit.This task clearly requires identification of all of the elements.To assess whether processing was serial or parallel, they didnot simply vary the number of stimuli in the display, they alsomanipulated the quality of the display. That is, on some trialsthe digits were bright and on others they were dim. The logicof this experimental paradigm, introduced by Sternberg(1967), is as follows: Let us suppose that a dim digit requiresk ms longer to encode than does a bright digit. If the subjectperforms the task by serially encoding each item in the dis-play, then the reaction time to a dim display with d digitsshould take kd ms longer than if the same display were bright.In other words, the effect of display size should interact mul-tiplicatively with the visual quality manipulation. However,if encoding of all the digits takes place simultaneously, thenthe k ms should be added in just once regardless of displaysize. In other words, display size and visual quality should beadditive. It was this latter effect that Pashler and Badgio ac-tually observed in their experiment, suggesting that the iden-tities of several digits could be accessed in parallel.

Attention: Types and Tokens

Recently, the notion that attention acts on the outputs of earlyfilters dedicated to processing simple features such as mo-tion, color, and orientation has been challenged by the ideathat the units on which attention operates are temporarystructures stored in a capacity-limited store usually referred

286 Attention

to as the visual short-term memory (vSTM). Several authorshave invoked the existence of such structures in the last10 years or so.

Pylyshyn and Storm (1988) proposed the notion of fingersof instantiation (FINSTs), which are similar to Marr’s (1983)place tokens because they represent filled locations indepen-dently of the features they contain. FINSTs provide accesspaths to attended objects, and we can monitor only a limitednumber of them (about five) simultaneously as they moveacross the visual field.

Kahneman et al. (1992) established a distinction betweenrepresentations stored in long-term memory, which are used inidentifying and classifying objects, and temporary episodicrepresentations called object-files (see also Kahneman &Treisman, 1984). An object-file is a spatiotemporal structurein which the information about a particular object is stored andcontinually updated. Consequently, an object, the variousproperties of which change over time, retains its identity solong as the information about its successive states is assignedto the same temporary object-file. When the changes are largeenough to disrupt the object’s spatiotemporal continuity, anew object-file is set up. According to the theory, the informa-tion contained in an object-file becomes available when atten-tion is allocated to it. Borrowing from this notion, Wolfe andBennett (1997) suggested that preattentive object-files areloose collections of basic features, with focused attentionneeded to appreciate the relationships among features.

The distinction between types and tokens later proposedby Kanwisher (e.g., Kanwisher, 1987; Kanwisher & Driver,1992) is essentially similar to Kahneman and Treisman’s(1984) distinction between nodes stored in a long-termrecognition network and temporary object-files, respectively.Kanwisher suggested that the activation of visual types andthe processing of spatiotemporal token information are inde-pendent processes performed in parallel, and that attention isrequired to integrate the information they provide aboutevents occurring in the visual field over space and time.

Finally, Rensink and colleagues (e.g., Rensink, 2000;Rensink et al., 1997) suggested that prior to focused atten-tion, low-level proto-objects are formed in parallel across thevisual field. Proto-objects are fairly complex preattentiverepresentations with limited spatiotemporal coherence, andas such, they are inherently volatile. Unless a proto-object be-comes the focus of attention, it is easily overwritten by astimulus that subsequently occupies its location or disinte-grates within a few hundred milliseconds, losing its continu-ity over time.

Although the various conceptualizations described abovemay differ along important aspects, they share a number ofcommon assumptions, namely, that (a) the visual system

establishes continuously changing temporary representa-tions (FINSTs, object-files, object tokens, or proto-objects);(b) these episodic representations should be distinguishedfrom properties such as color or shape that define an object’sidentity for categorization purposes; and (c) they require fo-cused attention in order to acquire spatiotemporal continuityor mediate conscious report. These notions have helped shedlight on a number of phenomena that have aroused great inter-est in the field of attention research in the last 10 years.

The Attentional Blink

In search experiments, even in the version in which subjectsmust identify all elements (e.g., the highest digit task), sub-jects typically must report only a single target on a trial. Dowe have the capacity to report several targets when those tar-gets are presented simultaneously or in temporal proximity?Duncan (1980) used the simultaneous-successive version ofthe visual search task described earlier. On each trial, fourcharacters were shown at the ends of an imaginary plus sign.The characters at 9:00 and 3:00 made up the horizontal limb,those at 12:00 and 6:00 the vertical limb. The displays con-sisted of digit targets and letter nontargets. The occurrence oftargets in the two limbs was independent. Thus, on a trialthere might be a target in one or the other limb or in bothlimbs (however, there was never more than one target in agiven limb). In the successive condition the two characters inone limb appeared briefly and were then masked; 500 mslater the two characters from the other limb were presentedbriefly and then masked. When only a single target was pre-sent on a trial there was no advantage for the successive-presentation condition. However, when there were twotargets present, accuracy in the simultaneous condition wassignificantly worse than in the successive condition. Thisdecrement cannot be attributed to the need to make two sep-arate overt responses; when subjects simply had to countthe number of targets (one vs. two targets present), the ad-vantage in the successive condition remained. Note also thatthe same results were obtained when a simple orientation dis-crimination was required to find the targets.

Recently, an interesting extension of this double-detectiontask has been explored intensively and has provided new in-sights into what mechanisms may underlie the limits revealedby double-detection experiments. It turns out that after a sub-ject has identified one target, it takes a surprisingly long timefor the system to recover to the point that it can efficientlyidentify a second target (e.g., Broadbent & Broadbent, 1987;Weichselgartner & Sperling, 1987). This refractory periodhas been dubbed the attentional blink (Raymond, Shapiro, &Arnell, 1992) or attentional dwell time (Duncan et al., 1994).

Closing Comments 287

In a typical attentional blink experiment, subjects are pre-sented with an RSVP stream of stimuli displayed sequentiallyat fixation at a rate of about 10 per sec. They are required torespond to two targets (by detecting or identifying them, de-pending on the studies). When the SOA (or lag) betweenthese targets ranges between 100 and 500 ms, performanceon the second target, given that the first target was correctlyreported, is severely impaired relative to that in a control con-dition in which the first target is ignored. Performance may ormay not be spared at short SOAs, but in any event recovers toits baseline level at longer SOAs (see Figure 10.8). The at-tentional blink effect does not require that the targets be em-bedded in an RSVP stream. Duncan et al. (1994) obtained theeffect using only two masked targets appearing at differentspatial locations in the visual field.

Although the underlying mechanisms postulated byvarious researchers may differ (e.g., Chun & Potter, 1995;Raymond et al., 1992), the prevailing view is that the atten-tional blink phenomenon reveals the effects of insufficientattentional resources’ being allocated to the second target. Itis assumed that whereas the second target receives some ini-tial processing, it does not reach the state at which it can bereported accurately. Shapiro, Driver, Ward, and Sorensen(1997) used a variant of the attentional blink paradigm inwhich subjects had to report three targets rather than two.They showed that performance on the third target (presentedat an SOA long enough to allow recovery from the blink) wasfacilitated when it was semantically related to the secondtarget, although the latter was poorly reported. They con-cluded that the attentional blink may reveal a failure to ex-tract visual tokens, which mediate conscious perception, butnot visual types, the activation of which underlies the primingeffects found in their study (see also Chun, 1997, for the sim-ilar notion that the attentional blink may reflect a general lim-itation in the binding of correctly identified types to objecttokens). For a further and more general discussion of refrac-tory effects, see the chapter by Proctor and Vu in this volume.

Repetition Blindness

In typical repetition blindness experiments (e.g., Kanwisher,1987; Mozer, 1989), subjects are required to report two tar-gets embedded in an RSVP stream. Performance on the sec-ond target is worse when it is identical to the first target thanwhen it is different, even when the two targets are separatedby intervening stimuli. Similar results are obtained whenthe targets are presented simultaneously rather than sequen-tially (e.g., J. Kim & Kwak, 1990; Santee & Egeth, 1980).Kanwisher as well as several other investigators (e.g.,Hochhaus & Marohn, 1991; Mozer, 1989) accounted for this

phenomenon by proposing that the second occurrence of arepeated item is recognized as a visual type, but is not indi-viduated as a distinct event. In other words, repetition blind-ness is assumed to reflect a failure in token individuation. Inthe absence of a separate token providing the spatiotemporalinformation necessary to distinguish between successive ac-tivations of the same type, the percept of the second instancebecomes assimilated into the percept of the first instance.Consistent with this hypothesis, Chun (1997) showed thatenhancing the episodic distinctiveness of the two targets bypresenting them in different colors causes the repetitionblindness effect to disappear, at least when subjects are givenenough practice and learn to use the color cue.

The Aftermath of Attention

One might wonder about the aftermath of attention. If attend-ing to an object binds together its features and permits thedetection of a change in the object, for how long do thesebenefits last? Rensink (2000) suggests not very long. Basedon the assumption that only one object can be represented ata time, if attention is switched to another object, the previ-ously attended parts of the visual field revert to their originalstatus as volatile proto-objects. Wolfe, Klempen, and Dahlen(2000) used a standard visual search task in which subjectslooked for a target item among distractors. In one condition,a new search display was presented on each trial. In anothercondition, the same display was used repeatedly. The strikingresult was that search did not become more efficient withextensive use of the same display. Wolfe et al. concluded thatthe effects of attention have no cumulative effect on visualperception. As they put it, “attention to one object after an-other may cause an observer to learn what is in a visualdisplay, but it does not cause that observer to see the visualdisplay in any different manner” (p. 693). In short, to the ex-tent that preattentive vision consists of “shapeless bundles ofbasic features” (Wolfe & Bennett, 1997), then so does post-attentive vision.

CLOSING COMMENTS

In this chapter we have explored a large number of behav-ioral paradigms. We have considered what captures attention,and how attention behaves over space and time (and over ob-jects situated in space and time). Although much has beenlearned about attention in the past century, and although thepace of discovery is (if anything) accelerating, there aremany more questions that need to be answered. This reviewhas been necessarily brief. For a more complete discussion of

288 Attention

the topics covered in this chapter the reader is directed to thebooks by Pashler (1998) and van der Heijden (1992). It isworth noting explicitly that the present discussion has beenalmost entirely concerned with behavioral studies. We havebarely touched on some of the other approaches that havebeen taken to the study of attention. In particular, readerswishing to learn about the neural bases of attention, as un-covered through studies using single-cell recordings inawake, behaving monkeys, through brain imaging or evokedpotential studies of humans, or through the study of patientswith neuropsychological disorders, should consult the bookedited by Parasuraman (1998).

REFERENCES

Ashby, F. G., Prinzmetal, W., Ivry, R., & Maddox, W. T. (1996). Aformal theory of feature binding in object perception. Psycho-logical Review, 103, 165–192.

Bacon, W. F., & Egeth, H. E. (1994). Overriding stimulus-driven at-tentional capture. Perception & Psychophysics, 55, 485–496.

Bahcall, D. O., & Kowler, E. (1999). Attentional interference atsmall spatial separations. Vision Research, 39, 71–86.

Banks, W. P., & Prinzmetal, W. (1976). Configurational effects invisual information processing. Perception & Psychophysics, 19,361–367.

Baylis, G. C., & Driver, J. (1993). Visual attention and objects:Evidence for hierarchical coding of location. Journal of Experi-mental Psychology: Human Perception and Performance, 19,451–470.

Becker, L. (2001). Attending over space and objects. Unpublisheddoctoral dissertation, Johns Hopkins University, Baltimore, MD.

Becker, L., & Egeth, H. (2000). Mixed reference frames for dy-namic inhibition of return. Journal of Experimental Psychology:Human Perception and Performance, 26, 1167–1177.

Behrman, M., & Tipper, S. P. (1994). Object-based attentionalmechanisms: Evidence from patients with unilateral neglect.In L. Umilta & M. Moscovitch (Eds.), Attention and perfor-mance XV: Conscious and nonconscious information processing(pp. 351–375). Hillsdale, NJ: Erlbaum.

Ben-Av, M. B., Sagi, D., & Braun, J. (1992). Visual attention and per-ceptual grouping. Perception & Psychophysics, 52, 277–294.

Braun, J. (1998). Vision and attention: The role of training. Nature,393, 424–425.

Braun, J., & Sagi, D. (1990). Vision outside the focus of attention.Perception & Psychophysics, 52, 277–294.

Braun, J., & Sagi, D. (1991). Texture-based tasks are little affectedby second tasks requiring peripheral or central attentive fixation.Perception, 20, 483–500.

Bravo, M., & Blake, R. (1990). Preattentive vision and perceptualgroups. Perception, 19, 515–522.

Briand, K. A., and R. M. Klein (1987). Is Posner’s “beam” the sameas Treisman’s “glue”? On the relation between visual orientingand feature integration theory. Journal of Experimental Psychol-ogy: Human Perception and Performance, 13, 228–241.

Bridgeman, B., Hendry, D., & Stark, L. (1975). Failure to detect dis-placement of the visual world during saccadic eye movements.Vision Research, 15, 719–722.

Broadbent, D. E. (1958). Perception and communication. London:Pergamon Press.

Broadbent, D. E. (1982). Task combination and selective intake ofinformation. Acta Psychologica, 50, 253–290.

Broadbent, D. E., & Broadbent, M. H. (1987). From detection toidentification: Response to multiple targets in rapid serial visualpresentation. Perception & Psychophysics, 42, 105–113.

Butler, E., Mewhort, D. J., & Tramer, S. C. (1987). Location errorsin tachistoscopic recognition: Guesses, probe errors, or spatialconfusions? Canadian Journal of Experimental Psychology, 41,339–350.

Carter, R. C. (1982). Visual search with color. Journal of Experimen-tal Psychology: Human Perception and Performance, 8, 127–136.

Cave, K. R. (1999). The FeatureGate model of visual selection. Psy-chological Research, 62, 182–194.

Cave, K. R., & Bichot, N. P. (1999). Visuo-spatial attention:Beyond a spotlight model. Psychonomic Bulletin & Review, 6,204–223.

Cave, K. R., & Wolfe, J. M. (1990). Modeling the role of parallelprocessing in visual search. Cognitive Psychology, 22, 225–271.

Cheal, M., & M. Gregory (1997). Evidence of limited capacity andnoise reduction with single-element displays in the location-cuingparadigm. Journal of Experimental Psychology: Human Percep-tion and Performance, 23, 51–71.

Cherry, E. C. (1953). Some experiments on the recognition ofspeech with one and with two ears. Journal of the AcousticalSociety of America, 25, 975–979.

Chun, M. M. (1997). Temporal binding errors are redistributed bythe attentional blink. Perception & Psychophysics, 59, 1191–1199.

Chun, M. M., & M. C. Potter (1995). A two-stage model for multi-ple target detection in rapid serial visual presentation. Journal ofExperimental Psychology: Human Perception and Performance,21, 109–127.

Crick, F. H. C. (1984). The function of the thalamic reticular com-plex: The searchlight hypothesis. Proceedings of the NationalAcademy of Sciences USA, 81, 4586–4590.

Davis, G., Driver, J., Pavani, F., & Shepherd, A. (2000). Reapprais-ing the apparent costs of attending to two separate visual objects.Vision Research, 40, 1323–1332.

Deutsch, J. A., & Deutsch, D. (1963). Attention: Some theoreticalconsiderations. Psychological Review, 70, 80–90.

DiLollo, V. (1980). Temporal integration in visual memory. Journalof Experimental Psychology: General, 109, 75–97.

References 289

Donk, M. (1999). Illusory conjunctions are an illusion: The effectsof target-nontarget similarity on conjunction and feature errors.Journal of Experimental Psychology: Human Perception andPerformance, 25, 1207–1233.

Downing, C. J., & Pinker, S. (1985). The spatial structure of visualattention. In M. I. Posner & O. S. M. Marin (Eds.), Attention andperformance XI (pp. 171–187). Hillsdale, NJ: Erlbaum.

Duncan, J. (1980). The locus of interference in the perception of si-multaneous stimuli. Psychological Review, 87, 272–300.

Duncan, J. (1981). Directing attention in the visual field. Perception& Psychophysics, 30, 90–93.

Duncan, J. (1984). Selective attention and the organization of visualinformation. Journal of Experimental Psychology: General, 113,501–517.

Duncan, J. (1985). Visual search and visual attention. In M. I.Posner & O. S. M. Marin (Eds.), Attention and performance XI(pp. 85–105). Hillsdale, NJ: Erlbaum.

Duncan, J. (1987). Attention and reading: Wholes and parts inshape recognition—A tutorial review. In M. Coltheart (Ed.),Attention and performance XII: The psychology of reading(pp. 39–61). Hillsdale, NJ: Erlbaum.

Duncan, J., & Humphreys, G. W. (1989). Visual search and stimulussimilarity. Psychological Review, 96, 433–458.

Duncan, J., & Humphreys, G. W. (1992). Beyond the search surface:Visual search and attentional engagement. Journal of Experimen-tal Psychology: Human Perception and Performance, 18, 578–588.

Duncan, J., Ward, R., & Shapiro, K. L. (1994). Direct measurementof attentional dwell time in human vision. Nature, 369, 313–315.

Egeth, H., Jonides, J., & Wall, S. (1972). Parallel processing of mul-tielement displays. Cognitive Psychology, 3, 674–698.

Egeth, H. E., Virzi, R. A., & Garbart, H. (1984). Searching for con-junctively defined targets. Journal of Experimental Psychology:Human Perception and Performance, 10, 32–39.

Egly, R., Driver, J., & Rafal, R. D. (1994). Shifting visual atten-tion between objects and locations: Evidence from normal andparietal lesion subjects. Journal of Experimental Psychology:General, 123, 161–177.

Eriksen, C. W., & Hoffman, J. E. (1973). The extent of processing ofnoise elements during selective encoding from visual displays.Perception & Psychophysics, 14, 155–160.

Eriksen, C. W., & Yeh, Y. Y. (1985). Allocation of attention in thevisual field. Journal of Experimental Psychology: Human Per-ception and Performance, 11, 583–597.

Farah, M. J., Brunn, J. L., Wong, A. B., Wallace, M. A., & Carpenter,P. (1990). Frames of reference for allocating attention to space:Evidence from the neglect syndrome. Neuropsychologia, 28,335–347.

Fisher, D. L. (1984). Central capacity limits in consistent mapping,visual search tasks: Four channels or more? Cognitive Psychol-ogy, 16, 449–484.

Folk, C. L., Remington, R. W., & Johnston, J. C. (1992). Involuntarycovert orienting is contingent on attentional control settings.Journal of Experimental Psychology: Human Perception andPerformance, 18, 1030–1044.

Gatti, S. V., & Egeth, H. E. (1978). Failure of spatial selectivity invision. Bulletin of the Psychonomic Society, 11, 181–184.

Helmholtz, H. von. (1871). Ueber die Zeit welche notig ist, damitein Gesichtseindruck zum Bewusstsein kommt. BerlinerMonatsberichte, June 8, 333–337.

Henderson, J. M. (1991). Stimulus discrimination following covertattentional orienting to an exogenous cue. Journal of Experimen-tal Psychology: Human Perception and Performance, 17, 91–106.

Henderson, J. M. (1996). Spatial precues affect target discrimi-nation in the absence of visual noise. Journal of ExperimentalPsychology: Human Perception and Performance, 22, 780–787.

Hillstrom, A. P., & Yantis, S. (1994). Visual motion and attentionalcapture. Perception & Psychophysics, 55, 399–411.

Hillyard, S. A., & Munte, T. F. (1984). Selective attention to colorand location: An analysis with event-related brain potentials.Perception & Psychophysics, 36, 185–198.

Hochhaus, L. & Marohn, K. M. (1991). Factors in repetition blind-ness. Journal of Experimental Psychology: Human Perceptionand Performance, 17, 422–432.

Hoffman, J. E., & Nelson, B. (1981). Spatial selectivity in visualsearch. Perception & Psychophysics, 30, 283–290.

James, W. (1950). The principles of psychology (Vol. 1). New York:Dover. (Original work published 1890)

Jonides, J., & Mack, R. (1984). On the cost and benefit of cost andbenefit. Psychological Bulletin, 96, 29–44.

Jonides, J., & Yantis, S. (1988). Uniqueness of abrupt visual onsetin capturing attention. Perception & Psychophysics, 43, 346–354.

Joseph, J. S., Chun, M. M., & Nakayama, K. (1997). Attentionalrequirements in a preattentive feature search task. Nature, 387,805–807.

Kahneman, D., & Henik, A. (1981). Perceptual organization andattention. In M. Kubovy & J. R. Pomerantz (Eds.), Perceptualorganization (pp. 181–211). Hillsdale, NJ: Erlbaum.

Kahneman, D., & Treisman, A. (1984). Changing views on auto-maticity. In R. Parasuraman & R. Davies (Eds.), Varieties ofattention (pp. 29–62). New York: Academic Press.

Kahneman, D., Treisman, A., & Gibbs, B. J. (1992). The reviewingof object files: Object-specific integration of information. Cogni-tive Psychology, 24, 175–219.

Kanwisher, N. G. (1987). Repetition blindness: Type recognitionwithout token individuation. Cognition, 27, 117–143.

Kanwisher, N., & Driver, J. (1992). Objects, attributes, and visualattention: Which, what, and where. Current Directions in Psy-chological Science, 1, 26–31.

290 Attention

Kim, J., & Kwak, H. (1990). Stimulus repetition effects and thedimension-feature distinction in alternative targets. Journal ofExperimental Psychology: Human Perception and Performance,16, 857–868.

Kim, M. S., & Cave, K. R. (1995). Spatial attention in visual-searchfor features and feature conjunctions. Psychological Science, 6,376–380.

Kim, M. S., & Cave, K. R. (1999). Top-down and bottom-up atten-tional control: On the nature of the interference from a salientdistractor. Perception & Psychophysics, 61, 1009–1023.

Kleiss, J. A., & Lane, D. M. (1986). Locus and persistence of ca-pacity limitations in visual information processing. Journal ofExperimental Psychology: Human Perception and Performance,12, 200–210.

Koffka, K. (1935). Principles of Gestalt psychology. New York:Harcourt Brace.

Kramer, A. F., & Jacobson, A. (1991). Perceptual organization andfocused attention: The role of objects and proximity in visualprocessing. Perception & Psychophysics, 50, 267–284.

Kramer, A. F., Weber, T. A., & Watson, S. E. (1997). Object-basedattentional selection: Grouped arrays or spatially invariant repre-sentations?: Comment on Vecera and Farah (1994). Journal ofExperimental Psychology: General, 126, 3–13.

Kwak, H. W., Dagenbach, D., & Egeth, H. E. (1991). Furtherevidence for a time-independent shift of the focus of attention.Perception & Psychophysics, 49, 473–480.

LaBerge, D., & Brown, V. (1989). Theory of attentional operationsin shape identification. Psychological Review, 96, 101–124.

Lamy, D., & Egeth, H. E. (2002). Object-based selection: The role ofattentional shifts. Perception & Psychophysics, 64, 52–66.

Lamy, D., & Tsal, Y. (1999). A salient distractor does affect con-junction search. Psychonomic Bulletin & Review, 6, 93–98.

Lamy, D., & Tsal, Y. (2000). Object features, object locations andobject-files: Which does selective attention activate and when?Journal of Experimental Psychology: Human Perception andPerformance, 26, 1387–1400.

Lamy, D., & Tsal, Y. (2001). On the status of location in visual se-lective attention. European Journal of Psychology, 13(3), 305–342.

Lavie, N., & Tsal, Y. (1994). Perceptual load as a major determinantof the locus of selection in visual attention. Perception & Psy-chophysics, 56, 183–197.

Lu, Z. L., & Dosher, B. A. (2000). Spatial attention: Different mech-anisms for central and peripheral temporal precues? Journal ofExperimental Psychology: Human Perception and Performance,26, 1534–1548.

Luce,R.D. (1986).Response times:Their role in inferringelementarymental organization. New York: Oxford University Press.

Luck, S. J., & Thomas, S. J. (1999). What variety of attention is au-tomatically captured by peripheral cues? Perception & Psy-chophysics, 61, 1424–1435.

Mack,A., Tang, B., Tuma, R., Kahn, S., & Rock, I. (1992). Perceptualorganization and attention. Cognitive Psychology, 24, 475–501.

Marr, D. (1983). Vision. San Francisco: W. J. Freeman.

McElree, B., & Carrasco, M. (1999). The temporal dynamics of vi-sual search: Evidence for parallel processing in feature and con-junction searches. Journal of Experimental Psychology: HumanPerception and Performance, 25, 1517–1539.

McLean, J. P., Broadbent, D. E., & Broadbent, M. H. (1983). Com-bining attributes in rapid serial visual presentation tasks. Quar-terly Journal of Experimental Psychology: Human ExperimentalPsychology, 35A, 171–186.

Miller, J. (1982). Divided attention: Evidence for coactivation withredundant signals. Cognitive Psychology, 14, 247–249.

Moore, C. M., & Egeth, H. E. (1997). Perception without attention:Evidence of grouping under conditions of inattention. Journal ofExperimental Psychology: Human Perception and Performance,23, 339–352.

Moore,C.M.,&Egeth,H.E.(1998).Howdoesfeature-basedattentionaffect visual processing? Journal of Experimental Psychology:Human Perception and Performance, 24, 1296–1310.

Moore, C. M., Egeth, H. E., Berglan, L. R., & Luck, S. J. (1996). Areattentional dwell times inconsistent with serial visual search?Psychonomic Bulletin & Review, 3, 360–365.

Moore, C. M., Yantis, S., & Vaughan, B. (1998). Object-based vi-sual selection: Evidence from perceptual completion. Psycho-logical Science, 9, 104–110.

Moray, N. (1959). Attention in dichotic listening: Affective cues andthe influence of instructions. Quarterly Journal of ExperimentalPsychology, 11, 56–60.

Mozer, M. C. (1989). Types and tokens in visual letter perception.Journal of Experimental Psychology: Human Perception andPerformance, 15, 287–303.

Müller, H. J., & Rabbitt, P. M. A. (1989). Reflexive and voluntaryorienting of visual attention: Time course of activation andresistance to interruption. Journal of Experimental Psychology:Human Perception and Performance, 15, 315–330.

Murdock, B. B. (1971). A parallel-processing model for scanning.Perception & Psychophysics, 10, 289–291.

Näätänen, R. (1986). The neural-specificity theory of visual selec-tive attention evaluated: A commentary on Harter and Aine.Biological Psychology, 23, 281–295.

Nagy, A. L., & Sanchez, R. R. (1990). Critical color differences de-termined with a visual search task. Journal of the Optical Societyof America, 7, 1209–1217.

Nakayama, K., & Mackeben, M. (1989). Sustained and transientcomponents of focal visual attention. Vision Research, 29, 1631–1647.

Navon, D., & Ehrlich, E. (1995). Illusory conjunctions: Does inat-tention really matter? Cognitive Psychology, 29, 59–83.

Neisser, U. (1967). Cognitive psychology. New York: Appleton-Century-Crofts.

References 291

Palmer, J., Ames, C. T., & Lindsey, D. T. (1993). Measuring theeffect of attention on simple visual search. Journal of Experimen-tal Psychology: Human Perception and Performance, 19,108–130.

Palmer, S., & Rock, I. (1994). Rethinking perceptual organization:The role of uniform connectedness. Psychonomic Bulletin &Review, 1, 29–55.

Parasuraman, R. (Ed.). (1998). The attentive brain. Cambridge,MA: MIT Press.

Pashler, H. (1988a). Cross-dimensional interaction and texture seg-regation. Perception & Psychophysics, 43, 307–318.

Pashler, H. (1988b). Familiarity and visual change detection.Perception & Psychophysics, 44, 369–378.

Pashler, H. (1998). The psychology of attention. Cambridge, MA:MIT Press.

Pashler, H., & Badgio, P. C. (1985). Visual attention and stimulusidentification. Journal of Experimental Psychology: Human Per-ception and Performance, 11, 105–121.

Phillips, W. A. (1974). On the distinction between sensory storageand short-term visual memory. Perception & Psychophysics, 16,283–290.

Pollack, I. (1972). Detection of changes in spatial position: Short-term visual memory or motion perception. Perception & Psy-chophysics, 11, 17–27.

Posner, M. I. (1980). Orienting of attention. Quarterly Journal ofExperimental Psychology, 32, 3–25.

Posner, M. I., & Cohen, Y. A. (1984). Components of visual orient-ing. In H. Bouma & D. G. Bouwhuis (Eds.), Attention and per-formance X: Control of language processes (pp. 531–554).Hillsdale, NJ: Erlbaum.

Posner, M. I., Snyder, C. R., & Davidson, B. J. (1980). Attention andthe detection of signals. Journal of Experimental Psychology:General, 109, 160–174.

Prinzmetal, W., Presti, D., & Posner, M. I. (1986). Does attentionaffect visual feature integration? Journal of Experimental Psy-chology: Human Perception and Performance, 12, 361–369.

Pylyshyn, Z. W., & Storm, R. W. (1988). Tracking multiple inde-pendent targets: Evidence for a parallel tracking mechanism.Spatial Vision, 3, 179–197.

Raymond, J. E., Shapiro, K. L., & Arnell, K. M. (1992). Temporarysuppression of visual processing in an RSVP task: An attentionalblink? Journal of Experimental Psychology: Human Perceptionand Performance, 18, 849–860.

Rensink, R. A. (2000). The dynamic representation of scenes. VisualCognition, 7, 17–42.

Rensink, R. A., O’Regan, J. K., & Clark, J. J. (1997). To see or notto see: The need for attention to perceive changes in scenes.Psychological Science, 8, 368–373.

Rock, I., & Gutman, D. (1981). The effect of inattention on formperception. Journal of Experimental Psychology: Human Per-ception and Performance, 7, 275–285.

Rock, I., Linnett, C. M., Grant, P., & Mack, A. (1992). Perceptionwithout attention: Results of a new method. Cognitive Psychol-ogy, 24, 502–534.

Santee, J. L., & Egeth, H. E. (1980). Selective attention in thespeeded classification and comparison of multidimensional stim-uli. Perception & Psychophysics, 28, 191–204.

Schneider, W. X. (1993). Space-based visual attention models andobject selection: Constraints, problems and possible solutions.Psychological Research, 56, 35–43.

Shapiro, K., Driver, J., Ward, R., & Sorensen, R. E. (1997). Primingfrom the attentional blink: A failure to extract visual tokens butnot visual types. Psychological Science, 8, 95–100.

Shiffrin, R. M., & Gardner, G. T. (1972). Visual processing capacityand attentional control. Journal of Experimental Psychology, 93,72–82.

Shiu, L., & Pashler, H. (1994). Negligible effect of spatial precuingon identification of single digits. Journal of Experimental Psy-chology: Human Perception and Performance, 20, 1037–1054.

Simons, D. J. (2000). Current approaches to change blindness.Visual Cognition, 7, 1–15.

Simons, D. J., & Levin, D. T. (1998). Failure to detect changes topeople in a real-world interaction. Psychonomic Bulletin andReview, 5, 644–649.

Sperling, G., & Weichselgartner, E. (1995). Episodic theory of the dy-namics of spatial attention. Psychological Review, 102, 503–532.

Sternberg, S. (1967). Two operations in character recognition:Some evidence from reaction time measurements. Perception &Psychophysics, 2, 45–53.

Sternberg, S. (1969). The discovery of processing stages: Exten-sions of Donders’ method. In W. G. Koster (Ed.), Attention andPerformance II, Acta Psychologica, 30, 276–315. Amsterdam:North Holland.

Theeuwes, J. (1989). Effects of location and form cuing on the allo-cation of attention in the visual field. Acta Psychologica, 72,177–192.

Theeuwes, J. (1991). Cross-dimensional perceptual selectivity.Perception & Psychophysics, 50, 184–193.

Theeuwes, J. (1992). Perceptual selectivity for color and form.Perception & Psychophysics, 51, 599–606.

Theeuwes, J.,Atchley, P., & Kramer,A. F. (2000). On the time courseof top-down and bottom-up control of visual attention. In S.Monsell & J. Driver (Eds.), Attention and performance XVIII:Control of cognitive processes (pp.105–124). Cambridge, MA:MIT Press.

Theeuwes, J., & Burger, R. (1998). Attentional control during visualsearch: The effect of irrelevant singletons. Journal of Experi-mental Psychology: Human Perception and Performance, 24,1342–1353.

Tipper, S. P. (1985). The negative priming effect: Inhibitory primingby ignored objects. Quarterly Journal of Experimental Psychol-ogy: Human Experimental Psychology, 37A, 571–590.

292 Attention

Tipper, S. P., Brehaut, J. C., & Driver, J. (1990). Selection of mov-ing and static objects for the control of spatially directed action.Journal of Experimental Psychology: Human Perception andPerformance, 16, 492–504.

Tipper, S. P., Weaver, B., Jerreat, L. M., & Burak, A. L. (1994).Object-based and environment-based inhibition of return ofvisual attention. Journal of Experimental Psychology: HumanPerception and Performance, 20, 478–499.

Townsend, J. T. (1971). A note on the identifiability of parallel andserial processes. Perception & Psychophysics, 10, 161–163.

Townsend, J. T., & Ashby, F. G. (1983). The Stochastic modelingof elementary psychological processes. New York: CambridgeUniversity Press.

Treisman, A. (1960). Contextual cues in selective listening. Quar-terly Journal of Experimental Psychology, 12, 242–248.

Treisman, A. (1992). Spreading suppression or feature integration?A reply to Duncan and Humphreys (1992). Journal of Experi-mental Psychology: Human Perception and Performance, 18,589–593.

Treisman, A., & Gelade, G. (1980). A feature-integration theory ofattention. Cognitive Psychology, 12, 97–136.

Treisman, A., & Sato, S. (1990). Conjunction search revisited. Jour-nal of Experimental Psychology: Human Perception and Perfor-mance, 16, 459–478.

Treisman, A., & Schmidt, H. (1982). Illusory conjunctions in theperception of objects. Cognitive Psychology, 14, 107–141.

Trick, L. M., & Enns, J. T. (1997). Clusters precede shapes in per-ceptual organization. Psychological Science, 8, 124–129.

Tsal, Y. (1983). Movement of attention across the visual field.Journal of Experimental Psychology: Human Perception andPerformance, 9, 523–530.

Tsal, Y. (1989). Do illusory conjunctions support the feature inte-gration theory? A critical review of theory and findings. Journalof Experimental Psychology: Human Perception and Perfor-mance, 15, 394–400.

Tsal, Y., & Lamy, D. (2000). Attending to an object’s color entailsattending to its location: Support for location-special views ofvisual attention. Perception & Psychophysics, 62, 960–968.

Tsal, Y., & Lavie, N. (1988). Attending to color and shape: The spe-cial role of location in selective visual processing. Perception &Psychophysics, 44, 15–21.

Tsal, Y., & Lavie, N. (1993). Location dominance in attending tocolor and shape. Journal of Experimental Psychology: HumanPerception and Performance, 19, 131–139.

van der Heijden, A. H. C. (1992). Selective attention in vision. NewYork: Routledge.

van der Heijden, A. H. C. (1993). The role of position in object se-lection in vision. Psychological Research, 56, 44–58.

van der Heijden, A. H. C., Kurvink, A. G., de Lange, F., de Leeuw,F., & van der Geest, J. N. (1996). Attending to color with properfixation. Perception & Psychophysics, 58, 1224–1237.

Vecera, S. P. (1994). Grouped locations and object-based attention:Comment on Egly, Driver, and Rafal (1994). Journal of Experi-mental Psychology: General, 123, 316–320.

Vecera, S. P. (1997). Grouped arrays versus object-based represen-tations: Reply to Kramer, Weber, and Watson (1997). Journal ofExperimental Psychology: General, 126, 14–18.

Vecera, S. P., & Farah, M. J. (1994). Does visual attention selectobjects or locations? Journal of Experimental Psychology:General, 123, 146–160.

Virzi, R. A., & Egeth, H. E. (1984). Is meaning implicated in illu-sory conjunctions? Journal of Experimental Psychology: HumanPerception and Performance, 10, 573–580.

Visser, R. A. W., Bischof, W. F., & Di Lollo, V. (1999). Attentionalswitching in spatial and nonspatial domains: Evidence from theattentional blink. Psychological Bulletin, 125, 458–469.

von Wright, J. M. (1968). Selection in visual immediate memory.Quarterly Journal of Experimental Psychology, 20, 62–68.

Watson, S. E., & Kramer, A. F. (1999). Object-based visual select-ive attention and perceptual organization. Perception & Psy-chophysics, 61, 31–49.

Weichselgartner, E., & Sperling, G. (1987). Dynamics of automaticand controlled visual attention. Science, 238, 778–780.

Wolfe, J. M. (1994). Guided Search 2.0. A revised model of visualsearch. Psychonomic Bulletin & Review, 1, 202–238.

Wolfe, J. M. (1998). Visual search. In H. Pashler (Ed.), Attention(pp. 13–74). Hove, East Sussex, UK: Psychology Press Ltd.

Wolfe, J. M. (1997), & Bennett, S. C. (1997). Preattentive objectfiles: Shapeless bundles of basic features. Vision Research, 37,25–43.

Wolfe, J. M., Cave, K. R., & Franzel, S. L. (1989). Guided search:An alternative to the feature integration model for visual search.Journal of Experimental Psychology: Human Perception andPerformance, 15, 419–433.

Wolfe, J. M., Klempen, N. L., & Dahlen, K. A. (2000). Post-attentive vision. Journal of Experimental Psychology: HumanPerception and Performance, 26, 693–716.

Yantis, S. (1993). Stimulus-driven attentional capture and atten-tional control settings. Journal of Experimental Psychology:Human Perception and Performance, 19, 676–681.

Yantis, S., & Egeth, H. E. (1999). On the distinction between visualsalience and stimulus-driven attentional capture. Journal of Ex-perimental Psychology: Human Perception and Performance,25, 661–676.

Yantis, S., & Johnston, J. C. (1990). On the locus of visual selec-tion: Evidence from focused attention tasks. Journal of Experi-mental Psychology: Human Perception and Performance, 16,135–149.

Yantis, S., & Jonides, J. (1990). Abrupt visual onsets and selectiveattention: Voluntary vs. automatic allocation. Journal of Experi-mental Psychology: Human Perception and Performance, 16,121–134.

CHAPTER 11

Action Selection

ROBERT W. PROCTOR AND KIM-PHUONG L. VU

293

FUNDAMENTAL ISSUES, MODELS, AND THEORIES 294Historical Background 294Methodological and Modeling Issues 295Discrete and Continuous Models of Information

Processing 296Speed-Accuracy Trade-Off 297Psychophysiological Measures 297

RELEVANT STIMULUS INFORMATION 298Uncertainty and Number of Alternatives:

The Hick-Hyman Law 298Stimulus-Response Compatibility 299Sequential Effects 301Advance Information 303

RELEVANT AND IRRELEVANT STIMULUSINFORMATION 304Noncorrespondence of Relevant and

Irrelevant Information 304Negative Priming 306

MULTIPLE TASKS 307Task Switching 307Psychological Refractory Period 308Stop Signals 309

CHANGES IN ACTION SELECTION WITH PRACTICE 310APPLICATIONS 310SUMMING UP 312REFERENCES 312

Action selection refers to how a decision is made, typicallyunder speeded response conditions, regarding which of twoor more actions to take in response to perceptual events. It isusually studied using choice-reaction tasks in which subjectsmake assigned responses to stimuli as quickly and accuratelyas possible, and reaction time (RT) and response accuracy aremeasured. Action selection is often called response selection,but the term action selection has come to be used more fre-quently in recent years to emphasize that responses in choice-reaction tasks are goal-directed actions (Prinz, 1997).

A recent example of the importance of action selectionconcerns the notorious butterfly ballot used in Palm BeachCounty, Florida, for the 2000 U.S. presidential election. Theballot, shown in Figure 11.1, listed the names of candidates intwo columns, with the appropriate response being to insert astylus into a punch hole assigned to the candidate of choiceamong a centered column of holes. Although there was nofixed time limit for responding, the voters’ task was speededin the sense that a limited number of voting booths wereavailable, with many voters needing to use them. With thisballot, some voters apparently selected the second punch holeon the list, voting for Pat Buchanan, rather than the thirdpunch hole, which was assigned to Al Gore, for whom they

intended to vote. This selection error occurred because Gorewas listed in the second position of the left-hand column, im-mediately below the major opposing candidate, George W.Bush. Punch ballots most often list all candidates on the left-hand side, and their corresponding punch holes in the sameorder on the right. Because the relative location of Gore’s po-sition in the left-hand candidate list was second, previous ex-perience would lead voters to expect that the second holeshould be punched to vote for him. Moreover, this expectancyis consistent with the general principle that people tend tomake the response whose relative location corresponds to thatof the stimulus. Consequently, it is not surprising that somevoters would incorrectly punch the second hole instead of thethird one, even though arrows were used to mark the desig-nated punch holes for the candidates. The poor design of theballot caused a sufficient number of unintended votes forBuchanan, as well as discarded ballots for which the secondand third holes were punched, costing Gore the election.

As this example illustrates, the topic of action selection isundoubtedly important. However, action selection tends tobe viewed as peripheral to mainstream cognitive psychologyin the United States, as reflected in the fact that the topicis rarely mentioned in undergraduate cognitive psychology

294 Action Selection

texts. The view of many cognitive psychologists seems to bethat input and central processes can be investigated withoutone’s having to be concerned with the translation of the out-come of these processes into output. This view is ironic,given that a major impetus to the rise of contemporary cogni-tive psychology was research on human performance con-ducted by Paul Fitts (see Fitts & Posner, 1967), DonaldBroadbent (1958), and others in the 1950s. Outside of theUnited States, more recognition has been given to the impor-tance of selection and execution of action in human informa-tion processing. Action selection is seen as fundamentalbecause it involves the interface between perception and ac-tion. It is the theme of this chapter that action selection is ofvital importance to many of the phenomena studied in con-temporary cognitive psychology.

FUNDAMENTAL ISSUES, MODELS,AND THEORIES

Historical Background

Astronomers in the first half of the nineteenth century madethe initial contribution to the measurement of RT by estimat-

ing the time it took a star to reach the midline of a grid ofvertical lines relative to when it first entered the grid (seeWoodworth, 1938). Although this was a clever method ofmeasuring RT, individual differences in the judgment of whenthe star entered and reached the midline resulted in unreliablereadings from one astronomer to another. In an attempt tocompensate for individual differences, a personal equationwas developed in which a constant correction was made inorder to equate the readings of astronomers. However, laterinvestigations showed that the difference between two indi-viduals was not constant after all.

The study of action selection was of central concern inthe last half of the nineteenth century. Interest arose out ofissues concerning the speed of nerve transmission. Mostphysiologists thought that nerve transmission occurred toorapidly to be measured. However, Helmholtz (1850) con-ducted an experiment in which he stimulated motor nervesof frogs and measured the time between the presentation ofthe stimulus and muscular contraction. He estimated the rateof nerve transmission to be 26 m/s. One important contri-bution of this work was to demonstrate that the durationsof nervous systems’ processes are measurable. Helmholtzwas also the first to measure RT in a procedure intended to

Figure 11.1 Sample Palm Beach County, Florida, butterfly ballot in the 2000 U.S. presidential election.

Fundamental Issues, Models, and Theories 295

calculate the speed of nerve transmission in humans. Thisprocedure involved measuring RT as a function of the dis-tance away from the brain by applying a shock to the skin.However, Helmholtz concluded that this procedure does notyield an accurate measure of nerve conduction because themeasurements “suffer from the unfortunate fact that a partof the measured time depends on mental processes”(Helmholtz, 1867, p. 228).

The research of Helmholtz and others using RT to esti-mate the speed of nerve conduction stimulated Donders andhis students to pursue the use of RT as a means for measur-ing mental processes. De Jagger’s dissertation (1865/1970)provided the first account of the experiments conducted inDonders’s lab. The first part of De Jagger’s study continuedHelmholtz’s notion of measuring the speed of nerve conduc-tion, but the second part focused on measuring the time re-quired to identify a stimulus and select a motor response. Inone set of experiments, subjects were required to respond toa red light with the right hand and a white light with the lefthand. The mean RT was 356 ms, which was 172 ms longerthan a simple reaction (executing a single response when astimulus is presented) to the same stimuli. De Jagger inter-preted this time as the duration of the central processes in-volving stimulus discrimination and response initiation.

Donders (1868/1969) formalized the subtractive methodused by De Jagger, emphasizing specifically that the time fora particular process could be estimated by adding that processto a task and taking the difference in RT between the twotasks. He distinguished three types of reactions: type a (simplereaction), type b (choice reaction), and type c (go or no-goreaction; responding to one stimulus but not another). Thesetypes of reactions allowed separate measures of the stimulusidentification and decision processes that were assessedtogether by De Jagger. The difference between the type-cand type-a reactions was presumed to reflect the time forstimulus identification, and the difference between the type-band type-c reactions the time for “expression of the will”(p. 424).

Reaction time research in general, and the study of actionselection in particular, continued to flourish throughout theremainder of the nineteenth century (see Jastrow, 1890).Wundt (1883) criticized Donders for using the type-c reac-tion as a measure of stimulus identification, reasoning thatsubjects must distinguish whether to respond, and suggestedusing the type-d reaction instead as a pure measure. The type-d reaction is measured by presenting subjects with the samestimuli and having them make the same response every time,as in the type-a reaction, with the difference being that theyare instructed not to respond until they have identified thestimulus. However, Wundt’s type-d reaction quickly fell outof favor because it is subjective and highly variable, and after

practice, the type-d reaction time does not differ from thetype-a reaction time. Criticisms of the subtractive method ingeneral led to its demise in the early twentieth century.

Methodological and Modeling Issues

With the advent of the information processing approach inthe 1950s and 1960s, the subtractive method was resurrected.This method, and the stage analysis of RT data on which it isbased, came to be seen as sufficiently important to establishDonders as a major figure in the history of human perfor-mance. One influential use of the subtractive method was toestimate the rate of mental rotation by varying the amountthat one stimulus was rotated relative to another to whichit was to be compared, and measuring the slope of the RTfunction (Cooper & Shepard, 1973). Mean RT increased byapproximately 240 ms for each 20° increase in angle of rota-tion, suggesting a continuous transformation in which eachdegree of rotation took about 12 ms.

A major advance in stage analysis of RT data was the de-velopment of the additive factors method by Sternberg(1969). Like the subtractive method, the additive factorsmethod assumes discrete serial processing stages. However,whereas the subtractive method provides duration estimatesfor assumed stages, the additive factors method provides away to discover the stages themselves. Sternberg showedthat if two or more factors each influence the durations ofdistinct stages, then the effect of one of the factors on totalduration will be invariant across the levels of the otherfactors: That is, the effects of the variables on RT will beadditive. If two factors have interactive effects on RT,then they must influence at least one common stage. Thus,Sternberg advocated the use of multifactor experiments inwhich the presence or absence of interactions among vari-ables is used to determine the processing stages involved intask performance.

Numerous limitations of the additive factors method havebeen enunciated, including problems of accepting the nullhypothesis for additivity, assuming serial processing stageswith no feedback loops, and assuming constant output fromeach stage (see Pachella, 1974). Despite these limitations,the method has proven to be a useful tool for analyzingthe structure of information processing in a variety of tasks(see Sternberg, 1998) because, as Sanders (1998) states, “themethod appears to provide a successful summary of a largeamount of experimental data” (p. 65). One criterion for eval-uating the additive factors method is stage robustness: Therelations between two factors should not change as a functionof levels of other factors. Although there are exceptions,stage robustness has generally been found to hold (Sanders,1998).

296 Action Selection

RESPONSE EXECUTION RESPONSE EXECUTION

STIMULUSSTIMULUS

PROCESS 1

PROCESS 2

PROCESS 3

PROCESS 4

0 RTTIME

DISCRETE STAGES

0 RTTIME

PROCESSES IN CASCADE

PROCESS 1

PROCESS 2

PROCESS 3

PROCESS 4

Figure 11.2 Illustration of discrete stage model (left) and cascade model (right). Source: From McClelland (1979).

Discrete and Continuous Models ofInformation Processing

Sternberg’s (1969) additive factors method is based on a viewof human information processing that assumes that the pro-cessing sequence between stimulus and response consists ofa series of discrete stages, with each stage completing its pro-cessing before the next stage begins (see Figure 11.2, leftside). Other models allow for parallel or overlapping opera-tion of the different processing stages. McClelland (1979)proposed the cascade model of information processing inwhich partial information at one subprocess, or stage, istransferred to the next (see Figure 11.2, right side). Themodel assumes that each stage is continuously active and itsoutput is a continuous value that is always available to thenext stage. As in the discrete stage model, it is also assumedthat each stage operates only on the output from the preced-ing stage. The output of the final stage indicates which of thealternative responses to execute.

In the cascade model, an experimental manipulation mayaffect a stage by altering the rate of activation or the as-ymptotic level of activation. The asymptotic level is equiva-lent to the stage output in the discrete stage model, which isassumed to be constant, and the activation rate determines thespeed at which the final output is attained. Although the as-sumptions of the cascade model are different from those ofthe discrete stage model from which the additive factors

method was derived, the patterns of interactivity and additiv-ity can be interpreted similarly. For the cascade model, if twovariables affect the rate parameter of the same stage, their ef-fects on RT will be interactive; if each variable affects therate parameter of a different stage, their effects on RT will beadditive. In sum, as long as it is assumed that the final outputof a stage does not vary as a function of the manipulations,then use of the additive factors logic to interpret the RT pat-terns does not require an assumption of discrete stages.

Miller (1988) argued that the discrete versus continuouscategorization should not be viewed as dichotomous butas extremes on a quantitative dimension called grain size. Inhis words, “a variable is more continuous to the extent thatit has a small grain size and more discrete to the extent that ithas a large one” (p. 195). Miller suggested that there are threedifferent senses in which models of human information pro-cessing can be characterized as discrete or continuous: repre-sentation, transformation, and transmission.

Representation refers to the discrete/continuous nature ofthe input and output codes for the processing stage. For exam-ple, if the locations of stimuli and responses in two-choicespatial reaction tasks are coded as left or right in terms ofrelative position, as is often assumed, the spatial codes are dis-crete. However, if the locations are represented in terms of ab-solute positions in physical space, then the representations arecontinuous. Transformation refers to the nature of the opera-tion that the processing stage performs. The transformation of

Fundamental Issues, Models, and Theories 297

the stage that performs mental rotation is typically character-ized as continuous, in the sense that the mentally rotated ob-ject passes through a continuum of intermediate states from itsinitial orientation to the final orientation (Cooper & Shepard,1973). A completely discrete transformation would be to gen-erate the final orientation in a single step. For transmission, amodel is discrete if the processing of successive stages cannothave temporal overlap; that is, the next stage in the sequencemust wait until processing of the immediately precedingstage is completed. The discrete stage model underlyingSternberg’s (1969) additive factors method postulates discreterepresentation and transmission. McClelland’s (1979) cas-cade model, on the other hand, postulates continuous repre-sentation and transmission, as well as transformation.

A variety of models exist that are intermediate to thesetwo extremes. One such model is Miller’s (1982, 1988) asyn-chronous discrete coding model. This model assumes thatmost stimuli are composed of features, and these features areidentified separately. The processing is discrete in that eachfeature must be identified before output about it can bepassed to the response-selection stage. However, the identityof one feature may be passed to response selection whilestimulus identification processes are still operating on otherfeatures.

Speed-Accuracy Trade-Off

The subtractive and additive factors methods are usuallybased solely on RT data. However, RT in any specific task sit-uation is related to the number of errors that one is willing tomake. A person can respond rapidly and make many errors orslowly and make few errors. This relation is called the speed-accuracy trade-off, and the function plotting speed versusaccuracy is known as the speed-accuracy operating charac-teristic. For RT research, two aspects are crucial. First, ifslower RT is accompanied by lower error rate, then the RTdifference cannot be attributed unambiguously to differencesin processing efficiency. Second, under conditions in whichaccuracy is relatively high, as in most choice-reaction stud-ies, a small difference in error rate can translate into a largedifference in RT.

Because of this close relation between speed and accuracy,some researchers have advocated conducting experiments inwhich the speed-accuracy criterion is varied between blocksof trials (Dosher, 1979; Pachella, 1974). There are numerousways to vary the speed-accuracy criterion: payoffs, instruc-tions, deadlines, time bands (responding within a certain timeinterval), and response signals (responding when the re-sponse signal is presented; see Wickelgren, 1977, for details).When a speed-accuracy function is obtained, information is

provided about the intercept (time at which accuracy ex-ceeds chance), asymptote (the maximal accuracy), and rateof ascension from the intercept to the asymptote, each ofwhich may reflect different processes. Thus, a speed-accuracy study has the potential to be more informative thanone based solely on RT. However, speed-accuracy studiesrequire 5–10 times more data than RT studies and, in manycircumstances, do not provide better insight into the phenom-enon of interest.

In addition to looking at the macro trade-off producedby varying speed-accuracy emphasis across trial blocks, itis also possible to examine the micro trade-off between speedand accuracy of responding within a particular speed-accuracy emphasis block of the macro function. Models ofthe macro speed-accuracy trade-off can be differentiated ontheir predictions regarding the micro trade-off (Pachella,1974). Osman et al. (2000) presented strong empirical evi-dence that the macro and micro functions are independent. Intheir experiment, which used psychophysiological measuresas well as behavioral measures, the effect of the macro trade-off manipulation on RT was independent of that of the microtrade-off, with the micro trade-off affecting the part of the RTinterval prior to the lateralized readiness potential (an indica-tor of readiness to make a left or right response, describedlater) and the macro trade-off affecting the part of the RT in-terval after the lateralized readiness potential.

The best models currently for characterizing both RT andaccuracy data are sequential sampling models, which assumethat information gradually accumulates until a response crite-rion is reached (Van Zandt, Colonius, & Proctor, 2000). Inrandom walk models, a single counter records evidence asbeing toward one response criterion and away from another,or vice versa. In race models, separate counters accumulateevidence for each response alternative until the winnerreaches criterion. Sequential sampling models explain thespeed-accuracy trade-off by assuming that the response crite-ria are placed further from or closer to the starting point ofthe accumulation process. They explain biases toward oneresponse over another in terms of asymmetric settings of theresponse criteria for the respective alternatives. Althoughcontinuous models of this general type describe the relationbetween speed and accuracy well, discrete models that allowpure guesses on a certain percentage of trials can also explainthis relation.

Psychophysiological Measures

In recent years, there has been increasing use of psychophys-iological measures to supplement RT data (Rugg & Coles,1995). One of the most popular methods is to record

298 Action Selection

electroencephalograms (EEG), which measure voltagechanges in the brain over time from electrodes placed on thescalp. Of particular concern are event-related potentials(ERPs); these are voltage changes in the EEG elicited by aspecific event (e.g., a stimulus onset), averaged across manytrials to remove background EEG activities. One reason forthe popularity of ERPs is that, while a task is being per-formed, they provide continuous measures of brain activitypresumed to be systematically related to cognitive processes.By comparing the effects of task manipulations on variousERP components, their onset latencies, and their scalp distri-butions, one can make relatively detailed inferences about thecognitive processes. These inferences can be used, along withbehavioral measures, to evaluate alternative information pro-cessing models.

There are a number of different ERP components, or fea-tures, that are indicators of different aspects of processing.These are labeled according to their polarity, positive (P) ornegative (N), and their sequence or latency. Early compo-nents such as P1 and N1 (the first positive and negative com-ponents, respectively) are associated with early perceptualprocesses. They are called exogenous components becausethey occur in close temporal proximity to the stimulus eventand have a stable latency with respect to it. Later componentssuch as P3 (or P300) reflect cognitive processes such as at-tention. These components are called endogenous becausethey are a function of the task demands and have a more vari-able latency than the exogenous components. For example,when an occasional target stimulus is interspersed in a streamof standards, the P3 is observed in response to targets, but notto standards.

A measure that has been used extensively in studies of ac-tion selection is the lateralized readiness potential (LRP;Eimer, 1998), mentioned previously. This potential can berecorded in choice-reaction tasks that require a response withthe left or right hand. It is a measure of differential activationof the lateral motor areas of the visual cortex that occursshortly before and during execution of a response. The asym-metric activation favors the motor area contralateral to thehand making the response, because this is the area that con-trols the hand. Of importance, the LRP has been obtained insituations in which no overt response is ever executed, allow-ing it to be used as an index of covert, partial responseactivation. The LRP is thus a measure of the difference in ac-tivity from the two sides of the brain that can be used as an in-dicator of covert reaction tendencies, to determine whether aresponse has been prepared even when it is not actually exe-cuted. It can also be used to determine whether the effects ofa variable are prior or subsequent to response preparation,

as Osman et al. (2000) did. Falkenstein, Hohnsbein, andHoormann (1994) suggested that the latency of the LRP islinked most closely to central decision processes (i.e., actionselection), whereas the peak is more closely related to centralmotor processes.

Electrophysiological measurements and recordings ofmagnetic fields do not have the spatial resolution needed toprovide precise information about the brain structures thatproduce the recorded activity. Recently developed neuroimag-ing methods, including positron-emission tomography (PET)and functional magnetic resonance imaging (fMRI), measurechanges in blood flow associated with neuronal activity in dif-ferent regions of the brain. These methods have poor temporalresolution but much higher spatial resolution than the electro-physiological methods. Combined use of neuroimaging andelectrophysiological methods provides the greatest degree ofboth spatial and temporal resolution (Mangun, Hopfinger, &Heinze, 1998).

RELEVANT STIMULUS INFORMATION

Uncertainty and Number of Alternatives:The Hick-Hyman Law

Merkel (1885), described in Woodworth (1938), providedthe initial demonstration that RT increases as a functionof the number of possible alternatives. In Merkel’s experi-ment, the Arabic numerals 1–5 were assigned to the left handand the Roman numerals I–V to the right hand, in left-to-rightorder. Results showed that when the number of alternativesincreased from 2 to 10 choices, mean RT increased from ap-proximately 300 ms to a little over 600 ms.

Contemporary research dates from Hick’s (1952) andHyman’s (1953) studies in which the increase in RT withnumber of alternatives was tied to information theory,which quantifies information in terms of uncertainty (for Nequally likely alternatives, the number of bits of informa-tion is log2 N). The stimuli in Hick’s study were 10 lampsarranged in an irregular circle, and responses were 10 keyson which the fingers of the two hands were placed. InHyman’s study, the stimuli were eight lights correspondingto the eight corners of inner and outer squares, and eachlight was assigned a spoken name. In both studies, RT in-creased as a logarithmic function of the number of alterna-tives. Moreover, RT also varied systematically as a functionof the relative proportions of the stimulus-response (S-R)alternatives, the sequential dependencies, and speed-accuracy trade-off, as expected on the basis of informa-tion theory. This relation between RT and the stimulus

Relevant Stimulus Information 299

information that is transmitted in the responses is known asthe Hick-Hyman law:

RT � a � bHT,

where a is basic processing time and b is the amount that RTincreases with increases in the amount of information trans-mitted (HT; log2 N for equally likely S-R pairs with no errors).

The slope of the Hick-Hyman function is negativelycorrelated with measures of intelligence, which several re-searchers have claimed to reflect ability to process informa-tion rapidly (see Jensen, 1980). However, the fact that theslope of the function is highly dependent on the amount ofpractice (described later) and other factors severely limitsany conclusions that can be drawn from the negative correla-tion with intelligence tests. A recent study by Vickrey andNeuringer (2000) showed that the Hick-Hyman function hasa lower slope for pigeons than for humans, even when theyare tested in similar circumstances, which, if the relation tointelligence were accepted, would imply that pigeons aremore intelligent than humans.

One criticism of the Hick-Hyman law is that the functionrelating RT to number of alternatives is not logarithmic.Kvälseth (1980) introduced a variety of laws, including apower law for the case of equally likely alternatives and anexponential law for cases in which the alternatives are notequally probable. Longstreth, El-Zahhar, and Alcorn (1985)claimed that the specific power law, RT = a + b(1 – N–1),provides a better fit to data for equiprobable alternatives thanthe logarithmic function. Longstreth et al.’s main argumentfor the power law is that as the number alternatives increasesbeyond 8, the function is no longer linear with respect to thelogarithm, but becomes curvilinear (see Longstreth, 1988).Although theoretically derived from an attentional model,Longstreth et al.’s power law is a special case of the more gen-eral power law proposed by Kvälseth (1980). In addition,Kvälseth (1989) and Welford (1987) pointed out thatLongstreth et al.’s power law has several problems. Kvälseth(1989) captures the status of the Hick-Hyman law, stating,“Although, on purely empirical grounds, Hick-Hyman’s lawmay not be uniformly superior to other lawful relationships, ithas been clearly established that it does provide a good sum-mary description of a substantial amount of data” (p. 358).

Stimulus-Response Compatibility

Stimulus-response compatibility (SRC) is one of the princi-pal factors affecting efficiency of action selection. SRC refersto the fact that performance is better with some mappings of

stimuli to responses than with others. SRC effects are ubiqui-tous and occur with a variety of stimulus and response sets,although much of the research has focused on spatial SRCeffects.

Spatial Compatibility Effects

Paul Fitts is given credit for formalizing the concept ofSRC. Fitts and Seeger (1953) examined performance ofeight-choice tasks using all combinations of three stimulusarrangements and three response arrangements. They foundthat responses were faster and more accurate when the stim-ulus and response arrangements corresponded spatially thanwhen they did not. Fitts and Deininger (1954) showed that forconditions in which the stimulus and response arrangementswere the same, responses were much slower with an arbitrarymapping of S-R locations than with one in which the corre-sponding response was made to each stimulus. Even moreinteresting, performance was also much better with a mirror-reverse mapping of stimulus locations to response locationsthan with a random mapping, although performance was stillinferior to that of the spatially corresponding mapping.

The spatial SRC effect is robust in that it is obtained withauditory and tactual stimuli and with key presses, joystickmovements, and unimanual aimed movements (see Proctor &Reeve, 1990, and Hommel & Prinz, 1997, for edited volumeson SRC). The slope of the function for the Hick-Hyman law,relating RT to the number of alternatives, is inversely relatedto SRC (Smith, 1968), approaching zero for highly compati-ble S-R mappings (Teichner & Krebs, 1974). In other words,SRC effects increase in magnitude as the number of S-R al-ternatives increases.

Many studies have used a two-choice task in which a leftor right key press is made to a left or right stimulus. In two-choice tasks, responses are typically 50–100 ms faster whenthe S-R mapping is spatially compatible than when it is not,regardless of whether the stimuli are visual or auditory.Moreover, PET scans show increased bloodflow for incom-patible mappings compared to compatible mappings in thesame brain regions (left rostral dorsal premotor and posteriorparietal areas) for both visual and auditory modalities(Iacoboni, Woods, & Mazziotta, 1998). This spatial SRCeffect is a function of relative position of the stimuli andresponses: It occurs even when the stimulus display or handsare shifted to the left or right of center (Nicoletti, Anzola,Luppino, Rizzolatti, & Umiltà, 1982). Moreover, the SRCeffect is found when the hands are crossed so that the lefthand operates the right key and the right hand the left key(Roswarski & Proctor, 2000), as well as when the responses

300 Action Selection

are made with two fingers on the same hand (Heister,Schroeder-Heister, & Ehrenstein, 1990). The dependence ofthe effect on the spatial relations of the stimuli and responseshas led most accounts of spatial SRC to focus on spatial cod-ing as its basis. The spatial codes are based on the task goals,as illustrated in a study by Riggio, Gawryszewski, andUmiltà (1986) in which subjects operated the left key with astick held in the right hand and the right key with a stick heldin the left hand. Even though the hands were on their normalsides, responses were faster with the S-R mapping in whichthe stimuli corresponded to the location of the response keyand not the hand used for responding.

Conceptual, Perceptual, and Structural Similarity

A variety of SRC effects in addition to spatial compatibilityhave been demonstrated. Kornblum, Hasbroucq, and Osman(1990) and Kornblum and Lee (1995) have argued that SRCeffects will occur for any situation in which the stimulus andresponse sets have dimensional overlap (i.e., are similar).Dimensional overlap is presumed to include both conceptualand perceptual similarity. The role of conceptual similarity isillustrated in the findings that spatial SRC effects, broadly de-fined, occur when location words are spoken in response tophysical location stimuli, as well as when left-right keypresses are made to the words left and right or to left- andright-pointing arrows. The role of perceptual similarity isshown by the finding that SRC effects are larger within thespatial-manual and verbal-vocal modes, that is, for physicallocations mapped to key presses and location words mappedto naming responses, than between the modes (Wang &Proctor, 1996).

SRC effects are also obtained when the S-R sets do notshare conceptual or perceptual similarity but have structuralsimilarity. When an ordered set of stimuli (e.g., A, B, C, D) ismapped to an ordered set of responses (e.g., 1, 2, 3, 4), RT isshorter for a mapping that preserves or reverses this orderthan for one that does not. Another type of structural compat-ibility effect occurs when a symbolic two-dimensional stim-ulus set is mapped to index and middle finger responses oneach hand. When two letters (O, Z) of two sizes (large orsmall) are mapped to the responses, the left-to-right mappingof O, o, z, Z is easier than one of O, z, o, Z (Miller, 1982;Proctor & Reeve, 1985). Proctor and Reeve presented evi-dence that this difference is due to the letter identity distinc-tions being salient for the stimulus set and the distinctionsbetween the two left and two right responses being salient forthe response set. Performance is best for the condition inwhich the salient stimulus feature maps directly onto thesalient response feature. In other words, translation of the

specific stimulus into a response can occur more quicklywhen salient features correspond. Salient features coding hasbeen shown to determine the compatibility effects obtainedfor a variety of situations in which the stimulus and responsesets have structural similarity, but no conceptual or percep-tual similarity (Proctor & Reeve, 1990).

Compatibility Effects in Two Dimensions

Umiltà and Nicoletti (1990) examined compatibility along twodimensions in a two-choice task by varying the stimulus andresponse locations for a set of trials along a diagonal (see Fig-ure 11.3). They found that the compatibility effect was largerfor the horizontal dimension than for the vertical dimension, aphenomenon they called right-left prevalence. Vu and Proctor(2001) showed that this right-left prevalence effect can be re-versed to top-bottom prevalence by increasing the relativesalience of the vertical dimension. This was accomplished byusing response sets that emphasized the top-bottom distinction.In one experiment that showed top-bottom prevalence, subjectsresponded with anatomical top-bottom effectors, a hand andfoot. In another experiment, top-bottom prevalence was ob-tained when one hand was placed over the other so that the top-bottom distinction was salient. Thus, although right-leftprevalence typically is obtained when left-right effectors areused, and top-bottom prevalence when top-bottom effectors areused, the prevalence effects do not seem to have an anatomical

Figure 11.3 Illustration of the S-R compatibility conditions and subtasksin Umiltà and Nicoletti’s (1990) two-dimensional compatibility experi-ments. The stimuli (depicted by circles) and response keys (depicted bycylinders) were arranged along the same (bottom row) or different (top row)diagonals. By varying the mapping of stimuli to responses for each of thefour cells, mappings could be generated that were compatible on bothdimensions, compatible on the vertical dimension but not the horizontaldimension and vice versa, or incompatible on both dimensions.

Relevant Stimulus Information 301

basis, but are by-products of the relative salience of the twodimensions.

Compatibility effects can occur as well when the spatialdimension along which the stimulus locations vary is or-thogonal to that along which the response alternatives vary.For top-bottom stimuli mapped to left-right key press orvocal responses, the mapping of top-right and bottom-leftyields faster responding than the alternative mapping (Cho& Proctor, 2001). A variant of salient features coding canalso explain this mapping effect (Weeks & Proctor, 1990).Specifically, evidence indicates that the two alternatives onthe vertical and horizontal dimensions are coded asymmet-rically, with top and right being the polar referents for theirrespective dimensions. Consequently, the salient featurescoding explanation is that action selection occurs faster forthe top-right/bottom-left mapping than for the alternativemapping because it is the one for which the salient featurescorrespond. Adam, Boon, Paas, and Umiltà (1998) pro-posed that this asymmetric coding is a property of verbalcodes but not spatial codes. However, Cho and Proctorprovided evidence that it is a general property of spatialcoding.

With unimanual movements of a joystick or finger, thetop-right/bottom-left mapping is also typically more compat-ible than the alternative mapping. In this case, though, themapping preference is affected by the location of the re-sponse apparatus. The top-right/bottom-left advantage isenhanced when responding in the right hemispace, but it re-verses to a top-left/bottom-right advantage when respondingin the left hemispace (Weeks, Proctor, & Beyak, 1995). Lippa(1996) provided evidence that the mapping preference is alsoaffected by hand posture. According to her referential codinghypothesis, the finger-to-wrist axis provides a referenceframe that allows the response set to be coded parallel tothe stimulus set. For example, when left-right responses aremade with the right hand held at a comfortable 45–90º, theleft response can be coded as top and right response as bot-tom. Referential coding can explain many results obtainedwith unimanual responses, but it cannot explain why themapping preferences described above occur when thehand and finger are in a neutral posture that allows only left-right deflections perpendicular to the sagittal body midline(Michaels & Schilder, 1991).

Because of this deficiency of the referential coding hy-pothesis, Lippa and Adam (2001) proposed an end-state com-fort hypothesis. Similar to referential coding, the end-statecomfort hypothesis views orthogonal compatibility as a cor-respondence effect. However, it assumes that the responsedimension is mentally rotated, according to relative hand pos-ture, to bring it into alignment with the stimulus dimension.

The direction of rotation, clockwise or counterclockwise, isdetermined by physical constraints of the body. The responsedimension is mentally rotated in the direction that wouldyield the most comfortable end-state posture if the hand wereactually rotated (inward movement for the left or right handwhen positioned at centered or ipsilateral locations, and out-ward movement when positioned at contralateral locations).The end-state comfort hypothesis can account for more re-sults obtained with unimanual responses than the referentialcoding hypothesis, but both hypotheses are not directly ap-plicable to the orthogonal compatibility effects obtained withbimanual or vocal response sets.

Dual-Route Models

Virtually all explanations of SRC effects agree that at leastpart of the difference in RT between compatible and incom-patible mappings involves the time to translate the stimulusinto its assigned response based on the instructions providedfor the task. Translation is presumed to be fastest when anidentity rule can be applied (i.e., make the response corre-sponding to the stimulus), intermediate when some other rulecan be used (e.g., make the response that is the mirror oppo-site of the stimulus), and slowest when the response must beretrieved via the specific S-R associations defined for thetask. Although some models rely exclusively on intentionaltranslation (e.g., Rosenbloom & Newell, 1987), dual-routemodels that propose an additional direct (or automatic)response-selection route have come to be favored (e.g.,Kornblum et al., 1990; see Figure 11.4). The basic idea is thatwhen a stimulus occurs it tends to produce activation of itscorresponding response by way of long-term S-R associa-tions, regardless of the S-R mapping defined for the task. Theresulting activation produces a benefit in responding whenthe corresponding response is correct, but a cost when it isnot. The major reason that dual-route models have becomepopular is that correspondence effects often occur for irrele-vant stimulus dimensions (see Lu & Proctor, 1995), as dis-cussed in a subsequent section.

Sequential Effects

Repetition Benefit

Bertelson (1961) was the first to formally investigate sequen-tial effects on performance. He showed that for a two-choicetask, in which left-right stimuli were mapped compatibly toleft-right keys, the total response time for a set of trials wasless when the proportion of repetitions was .75 than whenit was .25. This repetition benefit was evident when the

302 Action Selection

response-stimulus interval (RSI) was 50 ms but not when itwas 500 ms.

Since Bertelson’s (1961) study, numerous, more detailedinvestigations of sequential effects in choice-reaction taskshave been conducted. First-order sequential effects are thosethat involve the relation of the current trial to the immediatelypreceding trial. The most common first-order effect is that theresponse to a stimulus is faster when the S-R pair for a trialis a repetition of the preceding S-R pair than when it is not.In two-choice tasks, this repetition benefit is obtained onlywhen the RSI is short. At RSIs of 500 ms or longer, a benefitfor alternations over repetitions is typically found instead.The repetition benefit is larger in tasks with more than twochoices, being an increasing function of the number of S-Ralternatives, and in these tasks a repetition benefit is foundeven at long RSIs (Soetens, 1998). The first-order sequentialeffects have been attributed to two processes, much likethose proposed for priming effects (Neely, 1977; chapter byMcNamara & Holbrook in this volume). At short RSIs, resid-ual activation from the preceding trial produces automatic fa-cilitation when the current trial is identical to it; at long RSIs,strategic expectancy regarding the nature of the next trial pro-duces faster responses for expected than unexpected stimuli(Soetens, 1998). This expectancy is for the alternative S-Rpair in two-choice tasks, but for repetition of the same pair intasks with more alternatives.

Pashler and Baylis (1991) evaluated the locus of the repe-tition benefit for tasks in which two stimuli were assigned toleft, middle, and right response keys operated by index, mid-dle, and ring fingers of the right hand. Two of the stimuliwere digits, two were letters, and two were nonalphanumeric

symbols (e.g., & and #). Stimuli were mapped to responsesin a categorizable (e.g., digits-to-left response, letters-to-middle response, and symbols-to-right response) or uncate-gorizable (e.g., a digit and a letter to the left response, etc.)manner. For both mappings, the repetition benefit occurredprimarily when the same stimulus was repeated and notwhen only the response was repeated. This repetition bene-fit for the same stimulus was not found when responses onalternate trials were vocal and manual. Consequently, Pashlerand Baylis concluded that the repetition effects were atthe stage of response selection, with the normal response-selection process being bypassed when the stimulus and re-sponse were repeated.

In Pashler and Baylis’s (1991) experiments, a benefit forresponse repetition alone tended to occur with categorizablebut not uncategorizable S-R mappings. Campbell and Proctor(1993) verified this effect, showing a benefit of approxi-mately 40 ms for response repetition alone with categorizablemappings but not uncategorizable mappings. Their remain-ing experiments showed that this response repetition benefit,as well as the additional benefit for repeating the same stimu-lus, could be obtained when the responses on successive tri-als were made with different hands. In the critical conditions,the stimuli were presented to the left or right of fixation onalternate trials, with responses to the left stimulus made withthe three fingers on the left hand and responses to the rightstimulus made with the three fingers on the right hand.A cross-hand repetition benefit was obtained when eitherspatial or finger information was consistent across hands,but not when both consistencies were eliminated. Theseresults imply that the response sets can be coded in terms of

StimulusElement

StimulusEncoding

rj Identity &Program

Sj

ResponseIdentification: rk Identity

RetrieveProgram

rk

?

-table lookup-search-rule-etc

Verificationrj = rk

Yes

CongruentExecute rj

Executerk

Abort rj

Incongruent

No

Figure 11.4 Illustration of the dimensional overlap model by Kornblum et al. (1990). The top route depicts automatic activation ofthe corresponding response, and the bottom route depicts identification of the assigned response by intentional S-R translation.Source: From Kornblum, Hasbroucq, & Osman (1990).

Relevant Stimulus Information 303

locations or effectors and that response selection benefitsfrom repetition of the stimulus category when it maps onto asalient feature of the response sets.

Soetens (1998) examined sequential effects for tasks inwhich subjects responded to four stimuli located at the cor-ners of an imaginary square by pressing the left key if thestimulus was to one side and the right key if it was to theother. When left-right stimulus locations were mapped com-patibly to left-right responses, the repetition benefit at theshort RSI (50 ms) was primarily associated with the response(i.e., the benefit was evident when the stimulus side wasthe same as on the previous trial, but the location was differ-ent). At the long RSI (1,000 ms), a small alternation benefitwas evident. With an incompatible S-R mapping (i.e., leftside to right response), the results were similar, but with anincreased benefit for repeating the same stimulus, particu-larly at the short RSI. When up-down responses were made tothe left-right stimulus locations, response and stimulus repe-tition benefits of similar magnitudes were found at the shortRSI. At the long RSI, the only effect was a repetition benefitfor the same stimulus. Soetens concluded that automatic fa-cilitation shifted toward stimulus-related processes as themapping became less compatible. Together, the studies ofPashler and Baylis (1991), Campbell and Proctor (1993), andSoetens indicate that response repetition, without stimulusrepetition, is beneficial when there is a structural relation be-tween the stimulus and response sets and that repetition of thestimulus is more important when the mapping is arbitrary.

Although first-order sequential effects have been mostwidely studied, second- and third-order repetition effects, in-volving the sequence of the preceding two or three stimuli,respectively, are larger and more consistent (Soetens, 1998).For two-choice tasks, at short RSIs, RT benefits from multi-ple repetitions, regardless of whether the present trial is arepetition or an alternation. For example, responses on thecurrent trial tend to be faster if the three preceding trials wererepetitions than if they were alternations. At long RSIs, how-ever, a prior string of repetition trials is beneficial if thecurrent trial is also a repetition, but a prior string of alterna-tion trials is beneficial if the current trial is an alternation.These two patterns of results can be attributed to automaticactivation and subjective expectancies, respectively. Thehigher order effects in Soeten’s study also showed the pat-terns indicative of automatic facilitation at the short RSI andsubjective expectancy at the long RSI.

Is the Hick-Hyman Law an Artifact of Repetition Effects?

Kornblum (1967, 1968) noted that, unless explicitly con-trolled, the proportion of repetition trials decreases as set size

increases. Therefore, he proposed that the Hick-Hyman lawis an artifact of repetition effects. Kornblum (1968) used afour-choice task in which four lights were mapped to four re-sponse keys and information was varied by manipulatingstimulus probabilities. For three levels of information, condi-tions were constructed in which the probability of repetitionwas high or low. RT was shorter for the high-repetition con-ditions than for the corresponding low-repetition conditions,and these latter conditions showed only a nonsignificant ef-fect of information on RT. Kornblum (1967) conducted asimilar experiment in which the number of alternatives wastwo, four, or eight. For four- and eight-choice tasks, RT wasshorter on repetition than on nonrepetition trials, with theslope being less for repetition trials. Within these tasks, RTfor repetition trials increased as the amount of stimulus infor-mation increased, but RT for nonrepetitions did not.

Hyman and Umiltà (1969) noted that the RSI inKornblum’s (1967, 1968) experiments was approximately140 ms, a short interval that would maximize repetition ef-fects and minimize preparation for the subsequent trial. Theyreplicated three of Kornblum’s (1968) conditions, but used anaverage RSI of 7.5 s. Although RT was faster for repetitionthan nonrepetition trials, the slopes of the two functions wereapproximately equal. Hyman and Umiltà concluded, “Thereseems little doubt that the information hypothesis is muchmore compatible with our results than those of Kornblum’s”(p. 47). In other words, the Hick-Hyman function is not anartifact of the proportion of repetition trials when there isadequate preparation time.

Advance Information

Warning Effects

Preparation is usually studied by presenting a neutral warningsignal at various intervals prior to the onset of the imperativestimulus. Bertelson (1967) had subjects press a right key to aright light and a left key to a left light. The warning signalwas an auditory click that, in different blocks, occurred 0, 20,50, 100, 150, 200, or 300 ms prior to the visual stimulus. Atthe 0-ms warning interval, RT was approximately 265 ms. Itdecreased to a minimum of 245 ms at the 150-ms interval andthen increased slightly to 250 ms at the two longest intervals.However, the error rate increased from about 7% at theshorter intervals to 12% at the 100- and 150-ms intervals, anddecreased slightly to 9% at the longer intervals. Thus, theeffect of the warning signal was to increase readiness torespond quickly, but at the expense of accuracy.

Posner, Klein, Summers, and Buggie (1973) obtained sim-ilar results for a two-choice task in which the compatibility of

304 Action Selection

the mapping of the stimulus locations to responses was ma-nipulated. Each trial was preceded by no warning or a 50-mswarning tone, followed at intervals of 50, 100, 200, 400, and800 ms by a stimulus to the left or right of fixation. RT was aU-shaped function of foreperiod, reaching a minimum at the200-ms interval. Error rate showed an opposing, invertedU-shaped function, being highest at the 100-ms interval. Themain effect of compatibility was significant in the RT anderror data, but compatibility did not interact with foreperiod.These results suggest that the warning tone altered alertness,or readiness to respond, but did not affect the rate at whichthe information built up in the response-selection system.

RT continues to increase as the foreperiod increases be-yond 800 ms, up to at least 5 s. Sanders and Wertheim (1973)failed to find an effect of foreperiod between 1 and 5 s for au-ditory stimuli, although they found the standard increase inRT for visual stimuli. However, Sanders (1975) demonstratedthat the critical factor seems to be stimulus intensity:Auditory stimuli below 70 dB showed foreperiod effectssimilar to those shown by visual stimuli, and there was atrend toward smaller effects for high-intensity visual signals.

Precuing Effects

Leonard (1958) was the first to demonstrate that subjects canuse advance information to prepare for a subset of S-R alter-natives. He tested himself in a six-choice reaction task inwhich six stimulus lights were mapped compatibly to six re-sponse keys pressed by the fingers of each hand. In the six-choice condition, all six stimuli were lit, and the target lightwent off 100 ms later. In a three-choice condition, only theleft or right set of three stimuli was used. Of most interestwas a precue condition in which the subject did not knowwhether the choice would involve the three left locations orthe three right locations until the lights designating those lo-cations were lit (i.e., those locations were precued). RT de-creased as a function of the precuing interval, with RT at the500-ms interval being equivalent to that of the three-choicetask.

Subsequent studies using four-choice tasks have obtainedsimilar results, in which the benefit for precuing the twoleft or two right locations occurs within the first 500 ms ofprecue onset (Miller, 1982; Reeve & Proctor, 1984). How-ever, when other pairs such as alternate locations are precued,the maximal benefit is not evident until a longer interval.Reeve and Proctor (1984) showed that the advantage for pre-cuing the two left or two right locations does not depend onthe fact that they typically involve responses from differenthands. With an overlapped hand placement in which the indexand middle fingers from the two hands are alternated, the twoleft or right locations show a similar precuing advantage

relative to other pairs of locations. These and other findingsimply that the time needed to obtain the maximal benefitfrom a precue varies as a function of how long it takes totranslate the precue information. Proctor and Reeve (1986)attributed this pattern of differential precuing benefits to thesalience of the left-right distinction.

Kantowitz and Sanders (1972) distinguished between twotypes of precue: utility and necessity. Utility precues, as in thestudies just discussed, are helpful in reducing the number ofalternatives, but do not provide information that is necessaryfor responding. Necessity precues tell subjects what informa-tion is relevant for the current trial (e.g., whether they are torespond to stimulus color or shape). RT is longer when theprecue is a necessity than when it is only useful. Because theinformation provided by necessity precues must be used at allintervals, it is more difficult to respond at shorter ISIs. Withutility precues, subjects use the information at longer inter-vals but not shorter ones.

RELEVANT AND IRRELEVANTSTIMULUS INFORMATION

Noncorrespondence of Relevant andIrrelevant Information

Effects of irrelevant information on performance have beenstudied extensively in many areas of experimental psy-chology. Three such effects studied in the choice reactionliterature—the Stroop color-naming effect, the Eriksenflanker effect, and the Simon effect—involve correspondenceof relevant and irrelevant stimulus information.

The Stroop Effect

The best-known example of irrelevant information affectingresponse selection is the Stroop color-naming task (seeMacLeod, 1991, for a review). In this task, color words arepresented in different ink colors, and subjects are instructedto name the ink color while ignoring the color word. InStroop’s (1935/1992) study, subjects took 110 s to name a listof 100 colors presented in incongruent color words, com-pared to 63 s to name a list of 100 colors presented in solidsquares. Thus, conflicting color words nearly doubled thenaming time, a phenomenon known as the Stroop effect.Stroop also reported that the time to read 100 color words inincongruent ink colors was 43 s, compared to 41 s when thewords were presented in black ink. Thus, the interferencewith color naming was asymmetric: Irrelevant words inter-fered with naming ink colors, but irrelevant ink colors did notinterfere with reading color words.

Relevant and Irrelevant Stimulus Information 305

This asymmetric pattern of interference has been reportedin numerous subsequent studies, including versions of thetask in which RTs to individual stimuli are recorded. An im-portant finding is that the pattern of asymmetry is dependenton the response mode. When the task involves pointing to amatching color, responses to color words are delayed by in-congruent colors, but responses to colors are not delayed byirrelevant color words (Durgin, 2000). Similarly, in spatialversions of the task, in which the word left or right is pre-sented in left or right locations or with an arrow pointing tothe left or right, the words produce interference when the re-sponses are made vocally, but the locations or arrows pro-duce interference when the responses are key presses (Lu &Proctor, 1995).

Stroop (1935/1992) showed in his Experiment 3 that a di-mension that does not produce interference (e.g., ink colorswhen the task is word reading) can be made to do so withpractice. In his experiment, subjects practiced four lists of50 words in the color-naming task for 8 days. The averagetime to read the list decreased from 50 s on the first day to33 s on the last day, but this was still longer than the 25 sto name a neutral list of colored swastikas. Subjects alsoperformed the word-reading task prior and subsequent topracticing the color-naming task. The time to perform theword-reading task was nearly twice as long (35 s) after thecolor-naming practice as before (19 s). Thus, the practice in-creased the strengths of the associations between colors andnames, and the colors now produced interference with read-ing color words.

More generally, relative strength of association is a goodpredictor of whether an irrelevant stimulus dimension willaffect responding to a relevant stimulus dimension. Lu andProctor (2001) classified the association of stimulus dimen-sions to key presses as high if they were both conceptuallyand perceptually similar (e.g., arrows are spatial and nonver-bal, as are key presses), intermediate if they were only con-ceptually similar (e.g., location words are spatial but verbal),and low if they were neither (e.g., colors and color words arenot similar to key presses). Across several experiments usingvarious combinations of relevant and irrelevant stimulusdimensions, the relative magnitudes of effect size werepredictable based on relative association strength. Baldo,Shimamura, and Prinzmetal (1998) obtained similar resultsvarying response modalities in addition to stimulus dimen-sions: Robust Stroop effects to location word/arrow stimuliwere observed when responding manually to location wordsor vocally to arrows, but not for the reverse relations. Theresults of Lu and Proctor and of Baldo et al. are generallyconsistent with Kornblum et al.’s (1990) emphasis on re-sponse activation varying as a function of dimensional over-lap and with parallel distributed processing models of the

type proposed by Cohen, Dunbar, and McClelland (1990),which rely on relative association strength.

The Eriksen Flanker Effect

Another widely studied effect of irrelevant information is theEriksen flanker effect (Eriksen & Eriksen, 1974). In the typi-cal experiment examining this effect, one or more stimuli areassigned to left-right responses. The target letter for each trialis presented at a known, centered location and is flanked byinstances of a distractor letter. In Eriksen and Eriksen’s ex-periment, the letters H and K were assigned to one responseand the letters S and C to the other response. The flankingletters could be the same as the target (HHHHHHH), the let-ter assigned to the same responses as the target (congruent;KKKHKKK), or a letter assigned to the opposite response(incongruent; SSSHSSS or CCCHCCC). When the letterswere in close spatial proximity, responses were faster whenthe flanking letters were identical to or congruent with thetarget than when they were incongruent. This congruencyeffect decreased as the spatial separation between the lettersincreased.

Because distractors that are not potential targets producelittle or no interference, the results suggest that the effectsreflect response activation. That is, the flanking letters acti-vate the response to which they are assigned, producingresponse competition when that response is not the one sig-naled by the target. This competition is evident in a tendencyfor the lateralized readiness potential to show initial activa-tion of the wrong response 150 to 250 ms after onset of thetarget and incongruent distractors (Gratton, Coles, Sirevaag,Eriksen, & Donchin, 1988). Eriksen and Schultz (1979) pro-posed a continuous flow account of the flanker effect, muchlike McClelland’s cascade model, in which stimulus informa-tion gradually accumulates in the visual system and continu-ously flows into the response system. Initially, a wide rangeof responses is activated, but as the output from the percep-tual system becomes more exact, the response activation be-comes increasingly restricted to the appropriate response.This account assumes that after a flanking letter is fully iden-tified, it will no longer produce response activation. How-ever, if it is assumed that fully identified flankers may stillcontribute to response activation, then discrete stage modelscan account for the results as well (Mordkoff, 1996).

The Simon Effect

The Simon effect is another close relative of the Stroop effect(Lu & Proctor, 1995). In the typical Simon task, stimulus lo-cation is irrelevant and the responses, most often left-rightkey presses, vary along a location dimension. The relevant

306 Action Selection

stimulus dimension typically involves a distinction other thanlocation (e.g., color or letter identity). The Simon effect isthat responses are faster when the location of the stimulusand response correspond than when they do not. The effecttypically is larger when responses are fast than when they areslow, implying that activation of the location information oc-curs quickly and then decreases because it is irrelevant to thetask (Hommel, 1993b). Consistent with this view, whenthe correct response is not the one that corresponds with thelocation of the stimulus, the lateralized readiness potentialshows evidence of slight, initial activation of the spatiallycorresponding response, which then shifts to activation of thecorrect, noncorresponding response (De Jong, Liang, &Lauber, 1994).

Considerable research on the Simon effect has focused onwhy stimulus location is coded when it is irrelevant to thetask. Stoffer and Umiltà (1997) attribute the Simon effect toshifts of attention associated with eye movements. Accordingto them, the position of the object attended at stimulus onset,typically a fixation point, provides a frame of reference. Thelocation of the stimulus relative to the focus of attention iscoded only when attention is shifted to the stimulus. Thiscode specifies the direction and amplitude of the saccade pro-gram to shift fixation to the stimulus. The types of evidencethey have presented in support of the attention-shifting hy-pothesis are that the Simon effect is absent when attentionshifts are prevented by the need to report a stimulus presentedat fixation and reversed when an attention shift back from thestimulus location to the fixation point is required.

Hommel (1993b) has argued instead that spatial codingoccurs with respect to various frames of reference, of whichthe focus of attention may be one. Perhaps the best evidencefor his referential coding hypothesis is that the Simon effectcan vary as a function of multiple frames of reference. In aprocedure used by Lamberts, Tavernier, and D’Ydewalle(1992) and Roswarski and Proctor (1996), a stimulus canoccur in one of eight locations, four to the left of fixation andfour to the right. Initially, four boxes appear to one or theother side to designate the possible locations for that trial.Then the two left or two right boxes disappear, and the im-perative stimulus is presented in one of the remaining boxes.In this case, a Simon effect occurs with respect to threeframes of reference: Left-right side of fixation; two left ver-sus two right on a side; and the left-right location within thefinal pair. The largest difference between corresponding andnoncorresponding responses occurs when the stimulus is inthe far left or far right location, for which all three spatialcodes are in agreement (e.g., all left or all right).

As with compatibility for relevant stimulus information,the Simon effect varies as a function of task goals. Hommel

(1993a) had subjects respond to a high or low pitch tone, pre-sented to the left or right side, by pressing a left or right key.The key closed a circuit that lit a light on the opposite side.When instructed to press the left key to the high pitch toneand the right key to the low pitch tone, a typical Simon effectoccurred. However, when instructed to turn on the rightlight to the high pitch tone and the left light to the low pitchtone, the Simon effect was a function of light location. Thatis, in this case, responses were faster when the stimulus wason the side opposite the responding hand, rather than on thesame side. Guiard (1983) obtained a similar finding in an ex-periment in which subjects responded to tone pitch by turninga steering wheel clockwise or counterclockwise. In the con-dition of most interest, the subject’s hands were placed at thebottom of the wheel, and a clockwise turn moved a cursorto a right target location and a counter-clockwise turn movedit to the left. Because of the hand placement, when thewheel was turned clockwise the hands moved to the left, andvice versa when the wheel was turned counter-clockwise.A Simon effect was obtained as a function of the directionof wheel rotation, rather than as a function of the directionin which the hands moved.

Another goal-related phenomenon is the Hedge andMarsh (1975) reversal, in which the Simon effect reverses tofavor noncorresponding locations when the response keys arelabeled according to the same dimension as the relevant stim-ulus information, and subjects are instructed to respond in anincompatible manner (e.g., press the green key to the redstimulus and vice versa). The explanation proposed by Hedgeand Marsh, and which has continued to be the most widelyaccepted, is that of logical recoding. The basic idea is thata respond opposite rule is applied both to the relevant stimu-lus dimension and, inadvertently, to the irrelevant locationdimension, leading to activation of the noncorrespondingresponse.

Negative Priming

For the Stroop color-naming task, and related tasks with ir-relevant stimulus information, the target stimulus value on atrial can not only be a repetition or nonrepetition of the rele-vant value on the previous trial, but also the same as the valueof the irrelevant information. When the value of the relevantstimulus dimension is the same as that of the irrelevant di-mension on the preceding trial, an effect called negativepriming is often observed. This effect was first demonstratedby Dalrymple-Alford and Budayr (1966) for the Stroopcolor-naming task. Subjects had to name the ink colors forlists of Stroop color words that differed in the relationbetween successive stimuli. For the control list, there was no

Multiple Tasks 307

relation between the word or ink color for successive stimuli.For the ignored repetition list, however, the irrelevant colorword for one stimulus was the relevant color for the nextstimulus. In other words, if the color word for stimulus n-1was red, then the ink color for stimulus n was red. The find-ing of interest was that the time to name the colors for the ig-nored repetition list was much longer than that for the controllist. This slowing of responses when the to-be-ignored infor-mation on the previous trial is relevant on the current trial isthe phenomenon of negative priming.

Negative priming has subsequently been studied mostoften using a method in which responses to individual stimuliare measured. In that situation, the trials are often presentedas pairs, with the first trial called a prime and the second aprobe. Negative priming is shown when responses are slowerfor trials in which the previously irrelevant information isnow relevant than for neutral trials. The negative priming ef-fect has been found in a variety of tasks for which irrelevantinformation is present (Fox, 1995; May, Kane, & Hasher,1995), including not only tasks that require identification ofan object but also those that require localization.

The most straightforward interpretation of negative prim-ing effects is that of selective inhibition: The irrelevant infor-mation must be inhibited in order to respond to the relevantinformation, and this inhibition carries forward to the nexttrial. Consequently, the response will be slowed if the inhib-ited information is now relevant. Although numerous find-ings are consistent with the selective inhibition hypothesis,they can also be accounted for without assuming inhibition.Moreover, the situation has been shown to be much morecomplex than the selective inhibition hypothesis suggests,and alternative explanations have been proposed. The twomost prominent alternatives are feature mismatching andepisodic retrieval. According to the feature mismatch hypoth-esis (Park & Kanwisher, 1994), symbol identities are boundto objects and locations, and any change in the bindings fromthe previous trial will produce negative priming. Theepisodic retrieval hypothesis (Neill & Valdes, 1992) statesthat presentation of a stimulus evokes retrieval of previousepisodes involving the stimulus. Because recent episodes aremost likely to be retrieved, if the target stimulus was a dis-tractor on the previous trial, the episode retrieved will includean ignore tag.

One problematic finding for the inhibition account is thatnegative priming effects do not appear to be short-lived.DeSchepper and Treisman (1996) found negative primingafter a delay of 30 days between the prime and probe trials. Inaddition, negative priming depends on the relation betweenthe prime and probe trials. For example, for the Stroop task,the effect is not found if the probe stimulus that follows the

prime Stroop stimulus is a color patch and not a colored word(Lowe, 1979). A simple inhibition account would seem topredict negative priming in this situation as well as in that forwhich the probe stimulus was a colored word.

MULTIPLE TASKS

Task Switching

In his classic monograph, “Mental Set and Shift,” Jersild(1927) began by saying, “The fact of mental set is primary inall conscious activity. The same stimulus may evoke any oneof a large number of responses depending upon the contex-tual setting in which it is placed” (p. 5). Jersild conductedexperiments in which subjects made a series of judgments re-garding each stimulus in a list as a function of whether asingle task was performed for all stimuli or two tasks wereperformed in alternating order. The major finding was that inmany situations the time to complete the list was longer formixed lists than for pure lists of a single task.

Beginning in the mid-1990s, there has been a resurgenceof interest in task switching. Research conducted on taskswitching, in which two tasks are presented in a fixed order(e.g., on alternate trials), has suggested that there are twocomponents associated with changing the task set from theprevious trial. One component involves voluntary prepara-tion for the forthcoming trial, with responses for the next trialbecoming progressively faster as the RSI increases. How-ever, time to prepare for the new task cannot be the only fac-tor contributing to the switching cost, because the cost is stillevident when the RSI is long (Allport, Styles, & Hsieh, 1994;Rogers & Monsell, 1995). A second component, whichAllport et al. (1994) called task set inertia and Rogers andMonsell (1995) called exogenous task set reconfiguration, isnot under the subject’s control. Apparently only a single trialwith the new task is necessary to complete configuration forthat task. Rogers and Monsell (Experiment 6) used sequencesof four task repetitions and then a switch to the alternate taskfor four consecutive trials, and so on, and found that theswitch costs were eliminated after the first trial of the newtask.

Shaffer (1965) conducted a study in which trials withcompatible and incompatible spatial mappings were ran-domly mixed. The stimulus to which the subject was torespond occurred in a left or right location, and a centeredhorizontal or vertical line signaled whether the mapping forthe trial was compatible or incompatible. When the mappingsignal occurred simultaneously with the stimulus, the stan-dard spatial compatibility effect was eliminated. Vu and

308 Action Selection

Proctor (2001) used stimulus color to designate the mappingand obtained similar results with left-right physical-locationstimuli, as Shaffer used, as well as with left-right pointing ar-rows. These findings are consistent with the fact that, in avariety of situations, performance of the easier of two tasksis harmed more by mixing (Los, 1996). However, Vu andProctor found that when the stimuli were the words left andright, the advantage for the compatible mapping was en-hanced compared to pure blocks of one trial type. These re-sults, along with many others, suggest that words areprocessed differently than physical locations and arrows.

Proctor and Vu (2002) also showed that mixing location-relevant and location-irrelevant trials within a trial blockalters the stimulus-response compatibility (SRC) effects ob-tained for each task. When physical location stimuli wereused to convey the location information, the standard SRCeffect was eliminated for location-relevant trials. However,the SRC effect was not affected with arrow stimuli and wasenhanced with location word stimuli. Mixing the two trialtypes also affects the Simon effect obtained for the location-irrelevant trials. For all stimulus types, when the location-relevant mapping was compatible, the Simon effect wasenhanced compared to pure blocks of Simon trials; whenthe location-relevant mapping was incompatible, a reverseSimon effect was obtained. With arrows and words, the re-verse effect was smaller than the positive effect. However,with physical locations, the reverse Simon effect was atleast as large as the positive effect obtained with the compat-ible location-relevant mapping. This outcome implies thatthere was no automatic activation of the corresponding re-sponse. The reversal for physical location stimuli obtainedwhen the location-relevant mapping was incompatible wasevident even when the trial type was precued by up to 2.4 sbefore presentation of the stimulus. This outcome indicatesthat the reversal does not reflect only a strategy of preparingthe noncorresponding response in anticipation that locationmay be relevant to the trial.

Psychological Refractory Period

In a common dual-task procedure, subjects perform twodifferent choice-reaction tasks, Task 1 (T1) and Task 2 (T2),on a single trial. The stimulus onset asynchrony (SOA) betweenthe stimuli for T1 (S1) and T2 (S2) is varied. The typical findingis that RT for the second task (RT2) is slowed as the SOA de-creases. Telford (1931) called this phenomenon the psycholog-ical refractory period (PRP) effect. Extensive research onthe PRP effect has been conducted over the past 50 years,and explanations have been proposed in terms of information-processing bottlenecks, demands on limited capacity resources,

and strategies adopted to satisfy task constraints (Meyer &Kieras, 1997; Pashler, 1998). The most widely accepted ac-count in recent years is a response-selection bottleneck modeladvocated by Pashler and colleagues (see Figure 11.5).Accord-ing to this model, stimulus identification and response execu-tion occur in parallel for the two tasks. However, responseselection operates serially because it requires a single-channelmechanism.

The evidence for the response-selection bottleneck modelcomes primarily from using locus of slack logic (Schweickert,1983) to interpret the patterns of additive and interactive ef-fects produced by variables presumed to selectively affectstimulus identification, response selection, and response exe-cution. According to the model, identification of S2 com-mences immediately upon its presentation, regardless of theSOA. At long SOAs response selection can begin as soon asstimulus identification is completed, but at short SOAs it can-not begin until response selection for T1 is finished. Conse-quently, there is slack in the processing sequence for T2between the completion of stimulus identification and initia-tion of response selection. At short SOAs, the slack can ab-sorb, at least in part, an increase in time to identify S2. Thisleads to the effect of the stimulus-difficulty manipulationbeing smaller at the short SOAs than at the long SOAs. In con-trast, for variables that affect response selection or responseexecution, which have their influence after the bottleneck,the extra time cannot be absorbed by the slack, and, therefore,their effects should be additive with those of SOA. These pre-dicted patterns of results have been found for several variablesof the respective types.

Meyer and Kieras (1997) have mounted a challenge tothe response-selection bottleneck model, arguing that evi-dence supporting it reflects a strategy adopted by subjectswhen the instructions state or imply that the response for T1

S1 R1

1A 1C1B

Time

2B 2C2A

S2 R2

Figure 11.5 Illustration of response selection bottleneck model. Stage A isstimulus identification, Stage B is response selection, and Stage C is re-sponse initiation. Response selection for Task 2 (Stage 2B) is delayed untilresponse selection for Task 1 (Stage 1B) is completed. S1 and R1 are thestimuli and responses for Task 1, and S2 and R2 are the stimuli and responsesfor Task 2.

Multiple Tasks 309

must be made before that for T2. They propose that there isno capacity limitation in processing other than a bottleneckfor response execution when the tasks require responses fromthe same output system (e.g., key presses for T1 and T2). Ac-cording to their strategic response deferment model, differentlock out strategies are adopted in specific situations to permitperformance of T1 and T2 in the manner requested. Whetherthe response-selection bottleneck is due to a structural limita-tion on information processing or a strategy adopted to sat-isfy task demands is an issue that remains to be resolved.

According to response-selection bottleneck accounts ofthe PRP effect, whether structural or strategic, response se-lection for T2 does not begin until that for T1 is completed.However, several recent studies have shown cross-talkeffects between T1 and T2 that imply that the T2 response isactivated before the response for T1 is selected. Hommel(1998) had subjects make a left or right key press to the colorof a red or green rectangle for T1 and say “red” or “green” tothe letter S or H for T2. RT for both tasks showed correspon-dence effects at short SOAs, with the response for each taskbeing faster when the color-naming response for T2 corre-sponded to the color for T1. Lien and Proctor (2000) obtainedsimilar results when T1 involved left-right key presses withthe left hand to low or high pitch tones and T2 left-right keypresses with the right hand to left-right arrow directions.Also, Logan and Schulkind (2000) reported correspondenceeffects for the categories of T1 and T2 stimuli for a variety oftasks. For example, when both tasks required letter-digit clas-sifications with left-right key presses on the left and righthands, respectively, RT was shorter when the two stimuliwere from the same category (e.g., letters) than when theywere not. The fact that, in all studies, the correspondence ef-fects are evident in RT1, as well as RT2, implies that the stim-ulus for T2 is translated into response activation prior to T1response selection. Hommel has proposed that such transla-tion of stimulus information into response activation is auto-matic, with the bottleneck being only in the final decisionabout which response to make for each task.

Stop Signals

A goal may change during the course of action selectionso that the action being selected is no longer relevant. Suchsituations have been studied in the stop-signal paradigm(Logan, 1994). In this paradigm, a choice-reaction task is ad-ministered, but a stop signal occurs at a variable interval afterthe imperative stimulus on occasional trials to indicate that aresponse should not be made. Of concern is whether thesubject is able to inhibit the response for the choice task.The response is more likely to be inhibited the shorter the

interval between the go and stop signals and the longer thechoice RT.

Performance on the stop-signal task has been interpretedin terms of a stochastic race model: The go process and stopprocess engage in a race. The response is executed if the goprocess finishes before the stop process and is inhibited if thestop process finishes first. This model predicts many featuresof the results obtained in the stop-signal task, including theprobability that the response will be inhibited as a function ofgo RT and stop-signal delay. The race model has been appliedto a variety of stimulus and response modes, suggesting thatit captures basic principles of action inhibition. However, itdoes not provide a detailed account of the processes underly-ing performance of specific tasks.

Logan and Irwin (2000) compared the processes involvedin inhibiting left-right key presses and left-right eye move-ments. Subjects responded to peripheral left-right stimuli orcentral left-right pointing brackets, with hand movements oreye movements, using a compatible or incompatible S-Rmapping. Estimates of stop-signal RT for hand movementswere similar for the two stimulus types and mappings. Stop-signal RT for eye movements was shorter than that for thehands, being shortest for the condition in which a compatiblemovement was made to a peripheral stimulus. These resultssuggest that the inhibition processes for hand and eye move-ments are different, although they follow the same basicprinciples.

Research has focused on trying to identify the point of noreturn, or the stage beyond which the response cannot bestopped. De Jong, Coles, Logan, and Gratton (1990) exam-ined this issue using left-right squeezing responses (to a cri-terion) to measure partial responses, the lateralized readinesspotential (LRP) to measure central response activation, andthe electromyogram (EMG) to measure muscle activation.LRP, EMG, and squeeze activity were found to occur onstop-signal trials for which the response was successfully in-hibited (i.e., did not reach criterion), which they interpretedas suggesting that no stage of response preparation is ballis-tic. However, Osman, Kornblum, and Meyer (1986) arguedthat the point of no return should be defined as the point atwhich the response cannot be stopped from beginning. Usingthis criterion, the partial squeezes in De Jong et al.’s study arecases of unsuccessful inhibition, indicating that muscle acti-vation is the point of no return. The evidence in De Jonget al.’s study favored two inhibitory mechanisms: Inhibitionof central activation processes was implicated because theLRP was truncated on successful stop trials, but several find-ings suggested that there was also a more peripheral mecha-nism of inhibition that affected the transmission of activationfrom central to peripheral structures.

310 Action Selection

CHANGES IN ACTION SELECTIONWITH PRACTICE

Choice RT decreases with practice at a task, with equivalentamounts of practice producing larger changes earlier in prac-tice than later. Teichner and Krebs (1974) reviewed numer-ous studies of visual choice reactions and concluded that thestage of processing that benefits most from practice is re-sponse selection. Newell and Rosenbloom (1981) proposedthat the changes in RT with practice follow a power function:

RT � BN��,

where N is the number of practice trials, B is RT on the firsttrial, and � is the learning rate. The power function has cometo be regarded as a law to which any model that is intended toexplain practice effects must conform. Although the powerlaw provides a good description of changes in RT withpractice averaged across subjects, Heathcote, Brown, andMewhort (2000) contend that it does not fit the functions forindividual performers adequately. They demonstrated thatexponential functions provided better fits than power func-tions to the data for individuals in 40 data sets, and proposeda new exponential law of practice.

Beginning with Merkel (1885), several investigators haveshown that the slope of the Hick-Hyman function decreaseswith practice (e.g., Hyman, 1953; Mowbray & Rhoades,1959). Seibel (1963) used all combinations of 10 lights as-signed directly to 10 keys. After more than 75,000 trials hadbeen performed, the RT for the 1,023-alternative task wasonly approximately 25 ms slower than that for a 31-alternativetask. Practice also is typically more beneficial for incompati-ble than compatible mappings. However, SRC effects do notdisappear even with considerable practice (Dutta & Proctor,1992; Fitts & Seeger, 1953).

Proctor and Dutta (1993) had subjects perform two-choicetasks for 10 blocks of 42 trials each. In the critical conditions,they performed with the hands uncrossed and crossed inalternate blocks. Whether compatible or incompatible, whenthe spatial mapping of left-right stimulus locations to left-right response locations remained constant, there was no costassociated with alternating the hand placements: Overall RTand changes with practice with the alternating placementswere comparable to those of subjects who practiced with thesame hand placement for all blocks. In contrast, when themapping of stimulus to response locations was switched be-tween blocks so that the same hand was used to respond to astimulus when the hands were crossed or uncrossed, therewas a substantial cost for participants who alternated handplacements compared to those who did not. These results

imply that the S-R associations that are strengthened throughpractice involve spatial response codes.

Practice with an incompatible spatial mapping alters theinfluence of stimulus location on performance when locationbecomes irrelevant to the task. Proctor and Lu (1999) hadsubjects perform a two-choice task for 3 days using an in-compatible spatial mapping. On the 4th day, they performeda task for which stimulus location was irrelevant. For thistask, the Simon effect was reversed, with RT faster for non-corresponding responses. Tagliabue, Zorzi, Umiltà, andBassignani (2000) found a similar effect of prior practicewith an incompatible mapping and showed it to be presenteven when subjects were tested a week later. Thus, a limitedamount of practice produces new spatial S-R associationsthat persist at a sufficient strength to override the preexistingassociations between corresponding locations.

Nissen and Bullemer (1987) demonstrated that when thetrials in a compatibly mapped four-choice spatial reaction taskfollow a sequence that repeats regularly (every 10 trials in theirstudy), performance improves more with practice than whenthe trial order is random. Considerable effort has been devotedsubsequently to determining whether this sequence learning isimplicit or explicit, and to examining the nature of whatis learned. Because this research is summarized in the chapterby Johnson in this volume, we restrict mention here to a studyby Koch and Hoffmann (2000).Across four experiments, theyvaried whether the stimuli were spatial or symbolic andwhether the responses were spatial or symbolic. Their resultsshowed that the effect of sequence repetition and structure onperformance was much stronger for spatial sequences thanfor symbolic sequences, regardless of whether the stimulus orresponse set was involved. Koch and Hoffmann also specu-lated that learning of the response sequence is greater for in-compatible S-R mappings (e.g., random mappings of digits toresponse locations) than for compatible S-R mappings. Re-gardless, they emphasized, “the selective impact of S-R com-patibility on learning stimulus and response sequences in SRT[serial reaction tasks] seems to us an important issue . . . thathas not received much attention” (p. 879).

APPLICATIONS

Contemporary research on action selection has its roots in dis-play-control design issues. Paul Fitts, who formalized the con-cept of SRC and conducted much of the groundbreakingresearch on action selection, was the founder of what is nowthe Fitts Human Engineering Division of theArmstrong Labo-ratory of the U.S. Air Force and made many contributions tohuman factors. Although most of the research on action

Applications 311

1

3

2

4

(a)

(b)

Figure 11.6 Illustration of various display-control mapping configurationsfor (a) stove tops (with specific burner-control pairings indicated by letters)and (b) Duncan’s (1977) four-choice tasks (with specific S-R pairings indi-cated by arrows).

selection has been basic in nature, the results obtained fromthis research are of considerable relevance to applications in-volving interface design. It is widely accepted that a user-friendly design must adhere to principles of action selection ingeneral and SRC in particular (see Andre & Wickens, 1990).

In a classic study of stove configurations, Chapanis andLindenbaum (1959) evaluated four control-burner arrange-ments (see Figure 11.6a). The experimenter demonstrated the

individual pairings of burners to controls for one of the fourstoves, and then instructed subjects to push the assigned con-trol to the burner that was lit. Subjects showed shorter RT forDesign 1 than Designs 2–4, for which RT did not differ ini-tially. Furthermore, no errors were made for the mappings ofDesign 1, whereas the overall error rate was 6%, 10%, and11% for Designs 2–4, respectively. Practice significantly re-duced RT and errors for Design 2 compared to Designs 3 and4, but performance was still worse than with Design 1. Whennaive subjects were asked which control-burner configura-tion was the best, most selected Design 1. However, theywere equally divided about whether Design 2, 3, or 4 wassecond best. Thus, although after practice performance wasbetter with Design 2 than Designs 3 and 4, naive subjects didnot anticipate this difference.

In a more recent study, Payne (1995) asked naive subjectsto rank from easiest to hardest the four mappings of a four-choice SRC task in which the inner or outer pairs are mappedcompatibly or incompatibly (Duncan, 1977; see Figure 11.6b).He compared the subjective rankings to RT measures obtainedby Duncan. Similar to the results of the stove study, subjectshad little difficulty identifying Design 4 as the easiest mappingbecause it was a direct mapping. However, more subjectsrated Design 1 (in which both inner and outer pairs weremapped incompatibly) as being harder than Designs 2 and 3(in which only one pair was mapped compatibly and the otherincompatibly), even though actual performance was secondbest on Design 1.

The deleterious effect of mixed mappings illustrated inDuncan’s (1977) study, as well as in that of Shaffer (1965),discussed in the “Task Switching” section, indicates that thecontext in which the display-control configuration is placedaffects performance. Most compatibility studies evaluate per-formance of specific S-R mappings in isolation. However,when more than one S-R mapping is used, the benefit that onemight expect from a compatible mapping is not always evi-dent. Moreover, if two tasks must be performed simultane-ously, it may be easier to perform them together when theyhave the same incompatible mapping than when one has acompatible mapping and the other does not (Duncan, 1979).Andre and Wickens (1990) refer to this benefit as one ofglobal consistency and note that it may sometimes be moreimportant than local compatibility.

Because the amount of experience with specific S-R map-pings is a major factor in efficiency of action selection, de-signers must take into account the stereotypic behavior of thespecific population for whom a product or system is beingdesigned. For example, from years of experience, Americansare more likely to flick a light switch down when they intendto turn a light off, whereas Englishmen are more likely to

312 Action Selection

flick the switch up. One of the most widely studied popula-tion stereotypes is that of direction of motion. Operators ofsystems that require control of direction of motion must makedecisions regarding which direction to move a control inorder to move the system or display indicator in a particulardirection. Across populations of individuals, many arrange-ments show preferred relations between the direction of con-trol action and outcome of system output (see Loveless,1962).

Obviously, when a linear control is in the same orientationas a linear display, the stereotype is to expect the display tomove in the same direction as the control. More interesting,when a linear display is oriented perpendicularly to the linearcontrol, right is paired with upward movement and left ispaired with downward movement. With control knobs, clock-wise rotation tends to be associated with up or right move-ments (see Hoffman, 1990a, 1990b). In addition, there is astereotype, called Warrick’s principle, that the display is ex-pected to move in the same direction as the part of the controlthat is nearest to the display. With a vertical display, a clock-wise rotation would be preferred if the control were locatedto the right of the display and a counter-clockwise rotation ifit were located to the left.

Hoffman (1990a, 1990b) evaluated the relative strength ofthe stereotypes for two- and three-dimensional display andcontrol relationships. He found that different populations (inthis case, engineers and psychologists) differed in their pref-erence. Engineers were more likely to follow Warrick’s prin-ciple, most likely because it has a mechanical basis, whereaspsychologists tended to follow the stereotype of preferringclockwise for up and right movement. This difference em-phasizes not only that the specific experience of individuals isimportant, but also that the preferred relations are based onthe individual’s mental model for the task.

SUMMING UP

Action selection is an important part of behavior inside andoutside of the laboratory because choices among alternativeactions are required in virtually all situations. Action selec-tion has been a topic of interest in human experimental psy-chology since Donders’s (1868/1969) seminal work, withcontemporary research on the topic being at the forefront ofthe cognitive revolution in the 1950s. S-R compatibility,which is the quintessential action-selection topic, saw a surgein research in the 1990s, with significant advances made inthe development of theoretical frameworks for explaining avariety of phenomena in terms of common mechanisms.As we move into the twenty-first century, the range of tasks

and environments in which compatibility effects play a sig-nificant role, and the significant insights these effects provideregarding human performance, is only now coming to befully appreciated.

REFERENCES

Adam, J. J., Boon, B., Paas, F. G. W. C., & Umiltà, C. (1998). Theup-right/down-left advantage for vertically oriented stimuli andhorizontally oriented responses: A dual-strategy hypothesis.Journal of Experiment Psychology: Human Perception andPerformance, 24, 1582–1597.

Allport, A., Styles, E. A., & Hsieh, H., (1994). Shifting intentionalset: Exploring the dynamic control of tasks. In C. Umiltà & M.Moscovitch (Eds.), Attention and performance: Vol. 15. Con-scious and nonconscious information processing (pp. 421–452).Cambridge, MA: MIT Press.

Andre, A. D., & Wickens, C. D. (1990). Display-control compat-ibility in the cockpit: Guidelines for display layout analysis.(ARL-90-12/NASA-A3I-90-1). Savoy, IL: University of Illinois,Aviation Research Laboratory.

Baldo, J. V., Shimamura, A., & Prinzmetal, W. (1998). Mappingsymbols to response modalities: Interference effects on Stroop-like tasks. Perception & Psychophysics, 60, 427–437.

Bertelson, P. (1961). Sequential redundancy and speed in a serialtwo-choice responding task. Quarterly Journal of ExperimentalPsychology, 13, 90–102.

Bertelson, P. (1967). The time course of preparation. QuarterlyJournal of Experimental Psychology, 19, 272–279.

Broadbent, D. E. (1958). Perception and communication. Oxford,England: Pergamon Press.

Campbell, K. C., & Proctor, R. W. (1993). Repetition effects withcategorizable stimulus and response sets. Journal of Experi-mental Psychology: Learning, Memory, and Cognition, 19,1345–1362.

Chapanis, A., & Lindenbaum, L. E. (1959). A reaction time study offour control-display linkages. Human Factors, 1(4), 1–7.

Cho, Y. S., & Proctor, R. W. (2001). Effect of an initiating action onthe up-right/down-left advantage for vertically arrayed stimuliand horizontally arrayed responses. Journal of ExperimentPsychology: Human Perception and Performance, 27, 472–484.

Cohen, J. D., Dunbar, K., & McClelland, J. L. (1990). On the con-trol of automatic processes: A parallel distributed processing ac-count of the Stroop effect. Psychological Review, 97, 332–361.

Cooper, L. A., & Shepard, R. N. (1973). Chronometric studies of therotation of mental images. In W. G. Chase (Ed.), Visual informa-tion processing (pp. 75–176). New York: Academic Press.

Dalrymple-Alford, E. C., & Budayr, B. (1966). Examination ofsome aspects if the Stroop color-word test. Perceptual andMotor Skills, 23, 1211–1214.

References 313

De Jagger, J. J. (1970). Reaction time and mental processes(J. Brozek & M. S. Sibinga, Translators). Nieuwkoop, TheNetherlands: B. De Graf. (Original work published 1865)

De Jong, R., Coles, M. G. H., Logan, G. D., & Gratton, G. (1990).Searching for the point of no return: The control of responseprocesses in speeded choice reaction performance. Journal ofExperimental Psychology: Human Perception and Performance,16, 164–182.

De Jong, R., Liang, C.-C., & Lauber, E. (1994). Conditional andunconditional automaticity: A dual-process model of effects ofspatial stimulus-response compatibility. Journal of ExperimentalPsychology: Human Perception and Performance, 20, 731–750.

DeSchepper, B., & Treisman, A. (1996). Visual memory for novelshapes: Implicit coding without attention. Journal of Experimen-tal Psychology: Learning, Memory, and Cognition, 22, 27–47.

Donders, F. C. (1969). On the speed of mental processes. In W. G.Koster (Ed.), Acta Psychologica, 30: Vol. 2. Attention and Per-formance (pp. 412–431). Amsterdam: North-Holland. (Originalwork published 1868)

Dosher, B. A. (1979). Empirical approaches to information process-ing: Speed-accuracy tradeoff functions or reaction time—Areply. Acta Psychologica, 43, 347–359.

Duncan, J. (1977). Response selection rules in spatial choice reac-tion tasks. In S. Dornic (Ed.). Attention and performance: Vol. 6(pp. 49–71). Hillsdale, NJ: Erlbaum.

Duncan, J. (1979). Divided attention: The whole is more than thesum of the parts. Journal of Experimental Psychology: HumanPerception and Performance, 5, 216–228.

Durgin, F. H. (2000). The reverse Stroop effect. PsychonomicBulletin & Review, 7, 121–125.

Dutta, A., & Proctor, R. W. (1992). Persistence of stimulus-responsecompatibility effects with extended practice. Journal of Exper-imental Psychology: Learning, Memory, and Cognition, 18,801–809.

Eimer, M. (1998). The lateralized readiness potential as an on-linemeasure of central response activation processes. BehaviorResearch Methods, Instruments, & Computers, 30, 146–156.

Eriksen, B. A., & Eriksen, C. W. (1974). Effects of noise letters uponthe identification of a target letter in a nonsearch task. Perception& Psychophysics, 16, 143–149.

Eriksen, C. W., & Schultz, C. W. (1979). Information processing invisual search: A continuous flow conception and experimentalresults. Perception & Psychophysics, 25, 249–263.

Falkenstein, M., Hohnsbein, J., & Hoormann, J. (1994). Effects ofchoice complexity on different subcomponents of the late posi-tive complex of the event-related potential. Electoencephalogra-phy and Clinical Neurophysiology, 92, 148–160.

Fitts, P. M., & Deininger, R. L. (1954). S-R compatibility: Corre-spondence among paired elements within stimulus and responsecodes. Journal of Experimental Psychology, 48, 483–492.

Fitts, P. M., & Posner, M. I. (1967). Human performance. Belmont,CA: Brooks/Cole.

Fitts, P. M., & Seeger, C. M. (1953). S-R compatibility: Spatialcharacteristics of stimulus and response codes. Journal ofExperimental Psychology, 46, 199–210.

Fox, E. (1995). Negative priming from ignored distractors invisual selection: A review. Psychonomic Bulletin and Review, 2,145–173.

Gratton, G., Coles, M. G. H., Sirevaag, E. J., Eriksen, C. W., &Donchin, E. (1988). Pre- and poststimulus activation of responsechannels:Apsychophysiological analysis. Journal of ExperimentalPsychology: Human Perception and Performance, 14, 331–344.

Guiard, Y. (1983). The lateral coding of rotations: A study of theSimon effect with wheel-rotation responses. Journal of MotorBehavior, 15, 331–342.

Heathcote, A., Brown, S., & Mewhort, D. J. K. (2000). The powerlaw repealed: The case for an exponential law of practice.Psychonomic Bulletin & Review, 7, 185–207.

Hedge, A., & Marsh, N. W. A. (1975). The effect of irrelevant spatialcorrespondences on two-choice reaction time. Acta Psycho-logica, 39, 427–439.

Heister, G., Schroeder-Heister, P., & Ehrenstein, W. H. (1990).Spatial coding and spatio-anatomical mapping: Evidence for ahierarchical model of spatial stimulus-response compatibility. InR. W. Proctor & T. G. Reeve (Eds.), Stimulus-response compati-bility: An integrated perspective (pp. 117–143). Amsterdam:North-Holland.

Helmholtz, H. (1850). Messungen über den zeitlichen Verlauf derZuckung animalischer Muskeln und die Fortpflanzungs-geschwindigkeit der Reizung in den Nerven [Measurement of thetime course of the muscles and the transmission speed of nervestimulation]. Archiv für Anatomie und Phsysiologie, 276–364.

Helmholtz, H. (1867). Mittheilung, betreffend Versuche über dieFortpflanzungsgeschwindigkeit der Reizung in den motorischenNerven des Menschen, welche Herr. N. Baxt aus Petersburg imPhysiologischen Laboratorium zu Heidelberg ausgeführt hat[Communication concerning experiments about the transmissionspeed of stimulation in the motor nerves of humans, whichMr. N. Baxt from Petersburg in the Physiology Laboratories atHeidelberg has performed]. Monatsberichte der Akademie derWissenschaften zu Berlin, 29, 228–234.

Hick, W. E. (1952). On the rate of gain of information. QuarterlyJournal of Experimental Psychology, 4, 11–26.

Hoffman, E. R. (1990a). Strength of component principles deter-mining direction-of-turn stereotypes for horizontally movingdisplays. Proceedings of the Human Factors Society 34th AnnualMeeting, 457–461.

Hoffman, E. R. (1990b). Strength of component principles fordirection-of-turn stereotypes of three-dimensional display/controlarrangements. Proceedings of the Human Factors Society 34thAnnual Meeting, 462–466.

Hommel, B. (1993a). Inverting the Simon effect with intention:Determinants of detection and extent of effects of irrelevant spa-tial information. Psychological Research, 55, 270–279.

314 Action Selection

Hommel, B. (1993b). The role of attention for the Simon effect.Psychological Research, 55, 208–222.

Hommel, B. (1998). Automatic stimulus-response translation indual-task performance. Journal of Experimental Psychology:Human Perception and Performance, 24, 1368–1384.

Hommel, B., & Prinz, W. (Eds.). (1997). Theoretical issues instimulus-response compatibility. Amsterdam: North-Holland.

Hyman, R. (1953). Stimulus information as a determinant of reac-tion time. Journal of Experimental Psychology, 45, 188–196.

Hyman, R., & Umiltà, C. (1969). The information hypothesis andnon-repetitions. Acta Psychologica, 30, 37–53.

Iacoboni, M., Woods, R. P., & Mazziotta, J. C. (1998). Bimodal(auditory and visual) left frontoparietal circuitry for senso-rimotor integration and sensoimotor learning. Brain, 121,2135–2143.

Jastrow, J. (1890). The time-relations of mental phenomena. NewYork: N.D.C. Hodges.

Jensen, A. R. (1980). Chronometric analysis of intelligence. Journalof Social Biological Structure, 3, 103–122.

Jersild, A. T. (1927). Mental set and shift. Archives of Psychology,14, (Whole No. 89).

Kantowitz, B. H., & Sanders, M. S. (1972). Partial advance infor-mation and stimulus dimensionality. Journal of ExperimentalPsychology, 92, 412–418.

Koch, I., & Hoffmann, J. (2000). The role of stimulus-based andresponse-based spatial information on sequence learning. Jour-nal of Experimental Psychology: Learning, Memory, and Cogni-tion, 26, 863–882.

Kornblum, S. (1967). Choice reaction time for repetitions and non-repetitions. Acta Psychologica, 27, 178–187.

Kornblum, S. (1968). Serial-choice reaction time: Inadequacies ofthe information hypothesis. Science, 159, 432–434.

Kornblum, S., Hasbroucq, T., & Osman, A. (1990). Dimensionaloverlap: Cognitive basis for stimulus-response compatibility—Amodel and taxonomy. Psychological Review, 97, 253–270.

Kornblum, S., & Lee, J.-W. (1995). Stimulus-response compatibil-ity with relevant and irrelevant stimulus dimension that do anddo not overlap with the response. Journal of Experimental Psy-chology: Human Perception and Performance, 21, 855–875.

Kvälseth, T. O. (1980). An alternative to the Hick-Hyman’s andSternberg’s Law. Perceptual and Motor Skills, 50, 1281–1282.

Kvälseth, T. O. (1989). Longstreth et al.’s reaction time model:Some comments. Bulletin of the Psychonomic Society, 27,358–360.

Lamberts, K., Tavernier, G., & D’Ydewalle, G. (1992). Effects ofmultiple reference points in spatial stimulus-response compati-bility. Acta Psychologica, 79, 115–130.

Leonard, J. A. (1958). Partial advance information in a choice reac-tion task. British Journal of Psychology, 49, 89–96.

Lien, M.-C., & Proctor, R. W. (2000). Multiple spatial corre-spondence effects on dual-task performance. Journal of

Experimental Psychology: Human Perception and Performance,26, 1260–1280.

Lippa, Y. (1996). A referential-coding explanation for compatibilityeffects of perceptually orthogonal stimulus and response dimen-sions. Quarterly Journal of Experimental Psychology, 49A,950–971.

Lippa, Y., & Adam, J. J. (2001). An explanation of orthogonal S-Rcompatibility effects that vary with hand or response position:The end-state comfort hypothesis. Perception & Psychophysics,63, 156–174.

Logan, G. D. (1994). On the ability to inhibit thought and action:A users’ guide to the stop signal paradigm. In D. Dagenbach &T. H. Carr (Eds.), Inhibitory processes in attention, memory, andlanguage (pp. 189–239). San Diego, CA: Academic Press.

Logan, G. D., & Irwin, D. E. (2000). Don’t look! Don’t touch!Inhibitory control of hand and eye movements. PsychonomicBulletin & Review, 7, 107–112.

Logan, G. D., & Schulkind, M. D. (2000). Parallel memory retrievalin dual-task situations: Semantic memory. Journal of Experi-mental Psychology: Human Perception and Performance, 26,1072–1090.

Longsteth, L. E. (1988). Hick’s law: Its limit is 3 bits. Bulletin of thePsychonomic Society, 26, 8–10.

Longstreth, L. E., El-Zahhar, N., & Alcorn, M. B. (1985). Excep-tions to Hick’s Law: Explorations with a response duration mea-sure. Journal of Experimental Psychology: General, 114, 41–74.

Los, S. A. (1996). On the origin of mixing costs: Exploring infor-mation processing in pure and mixed blocks. Acta Psychologica,94, 145–188.

Loveless, N. E. (1962). Direction-of-motion stereotypes: A review.Ergonomics, 5, 357–383.

Lowe, D. G. (1979). Strategies, context and the mechanisms ofresponse inhibition. Memory & Cognition, 7, 382–389.

Lu, C.-H., & Proctor, R. W. (1995). The influence of irrelevant locationinformation on performance: A review of the Simon effect and spa-tial Stroop effects. Psychonomic Bulletin & Review, 2, 174–207.

Lu, C.-H., & Proctor, R. W. (2001). Influence of irrelevant informa-tion on human performance: Effects of S-R association strengthand relative timing. Quarterly Journal of Experimental Psychol-ogy, 54A, 95–136.

MacLeod, C. M. (1991). Half a century of research on the Stroop ef-fect:An integrative review. Psychological Bulletin, 109, 163–203.

Mangun, G. R., Hopfinger, J. B., & Heinze, H.-J. (1998). Integratingelectrophysiology and neuroimaging in the study of human cog-nition. Behavior Research Methods, Instruments, & Computers,30, 118–130.

May, C. P., Kane, M. J., & Hasher, L. (1995). Determinants of neg-ative priming. Psychological Bulletin, 118, 35–54.

McClelland, J. L. (1979). On the time relations of mental processes:A framework for analyzing processes in cascade. PsychologicalReview, 88, 375–407.

References 315

Merkel, J. (1885). Die zeitliche Verhaltnisse de Willenstatigkeit[The temporal relations of activities of the will]. PhilosophischeStudien, 2, 73–127.

Meyer, D. E., & Kieras, D. E. (1997). A computational theory ofexecutive cognitive processes and multiple-task performance:Part 2. Accounts of psychological refractory-period phenomena.Psychological Review, 104, 749–791.

Michaels, C. F., & Schilder, S. (1991). Stimulus-responsecompatibilities between vertically oriented stimuli and horizon-tally oriented responses. Perception & Psychophysics, 49,342–348.

Miller, J. (1982). Discrete versus continuous stage models of humaninformation processing: In search of partial output. Journal ofExperimental Psychology: Human Perception and Performance,8, 273–296.

Miller, J. (1988). Discrete and continuous models of humaninformation processing: Theoretical distinctions and empiricalresults. Acta Psychologica, 67, 191–257.

Mordkoff, J. T. (1996). Selective attention and internal constraints:There is more to the flanker effect than biased contingencies. InA. F. Kramer, M. G. H. Coles, & G. D. Logan (Eds.), Convergingoperations in the study of visual selective attention (pp. 483–502).Washington D C: American Psychological Association.

Mowbray, G. H., & Rhoades, M. V. (1959). On the reduction ofchoice reaction times with practice. Quarterly Journal ofPsychology, 11, 16–23.

Neely, J. H. (1977). Semantic priming and retrieval from lexi-cal memory: Roles of inhibitionless spreading activation andlimited-capacity attention. Journal of Experimental Psychology:General, 106, 226–254.

Neill, W. T., & Valdes, L. A. (1992). Persistence of negative prim-ing: Steady state or decay? Journal of Experimental Psychology:Learning, Memory, and Cognition, 18, 565–576.

Newell, A., & Rosenbloom, P. S. (1981). Mechanisms of skillacquisition and the law of practice. In J. R. Anderson (Ed.),Cognitive skills and their acquisition (pp. 1–55). Hillsdale, NJ:Erlbaum.

Nicoletti, R., Anzola, G. P., Luppino, G., Rizzolatti, G., & Umiltà,C. (1982). Spatial compatibility effects on the same side of thebody midline. Journal of Experimental Psychology: HumanPerception and Performance, 8, 644–673.

Nissen, M. J., & Bullemer, P. (1987). Attentional requirementsof learning: Evidence from performance measures. CognitivePsychology, 19, 1–32.

Osman, A., Kornblum, S., & Meyer, D. E. (1986). The point of noreturn in choice reaction time: Controlled and ballistic stages ofresponse preparation. Journal of Experimental Psychology:Human Perception and Performance, 12, 243–258.

Osman, A., Lou, L., Muller-Gethmann, H., Rinkenauer, G., Mattes,S., & Ulrich, R. (2000). Mechanisms of speed-accuracy tradeoff:Evidence from covert motor processes. Biological Psychology,51, 173–199.

Pachella, R. G. (1974). The interpretation of reaction time ininformation-processing research. In B. H. Kantowitz (Ed.),Human information processing: Tutorials in performance andcognition (pp. 41–82). Hillsdale, NJ: Erlbaum.

Park, J., & Kanwisher, N. (1994). Negative priming for spatial loca-tions: Identity mismatching, not distractor inhibition. Journal ofExperimental Psychology: Human Perception and Performance,20, 613–623.

Pashler, H. (1998). The psychology of attention. Cambridge, MA:MIT Press.

Pashler, H., & Baylis, G. (1991). Procedural learning: 2. Intertrialrepetition effects in speeded choice task. Journal of ExperimentalPsychology: Learning, Memory, and Cognition, 17, 33–48.

Payne, S. J. (1995). Naïve judgments of stimulus-response compat-ibility. Human Factors, 37, 495–506.

Posner, M. I., Klein, R., Summers, J., & Buggie, S. (1973). On theselection of signals. Memory & Cognition, 1, 2–12.

Prinz, W. (1997). Why Donders has led us astray. In B. Hommel &W. Prinz (Eds.), Theoretical issues in stimulus-response compat-ibility (pp. 247–267). Amsterdam: North-Holland.

Proctor, R. W., & Dutta, A. (1993). Do the same stimulus-responserelations influence choice reactions initially and after practice?Journal of Experimental Psychology: Learning, Memory, andCognition, 19, 922–930.

Proctor, R. W., & Lu, C.-H. (1999). Processing irrelevant locationinformation: Practice and transfer effects in choice-reactiontasks. Memory & Cognition, 27, 63–77.

Proctor, R. W., & Reeve, T. G. (1985). Compatibility effects in theassignment of symbolic stimuli to discrete finger response.Journal of Experimental Psychology: Human Perception andPerformance, 11, 623–639.

Proctor, R. W., & Reeve, T. G. (1986). Salient-feature codingoperations in spatial precuing tasks. Journal of Experimental Psy-chology: Human Perception & Performance, 12, 277–285.

Proctor, R. W., & Reeve, T. G. (Eds.). (1990). Stimulus-response com-patibility:An integrated perspective.Amsterdam: North-Holland.

Proctor,R.W.,&Vu,K.-P.L. (2002).Eliminating,magnifying,andre-versing spatial compatibility effects with mixed location-relevantand irrelevant trials. In W. Prinz & B. Hommel (Eds.), Commonmechanisms in perception and action: Attention and performance,XIX (pp. 443–473). NewYork: Oxford University Press.

Reeve, G. T., & Proctor, R. W. (1984). On the advance preparationof discrete finger responses. Journal of Experimental Psychol-ogy: Human Perception and Performance, 10, 541–553.

Riggio, L., Gawryszewski, L. G., & Umiltà, C. (1986). What iscrossed in crossed-hand effects? Acta Psychologica, 62, 89–100.

Rogers, R. D., & Monsell, S. (1995). Cost of a predictable switchbetween simple cognitive tasks. Journal of Experimental Psy-chology: General, 124, 207–231.

Rosenbloom, P. S., & Newell, A. (1987). An integrated computa-tional model of stimulus-response compatibility and practice. In

316 Action Selection

G. H. Bower (Ed.), The psychology of learning and motivation(Vol. 21, pp. 1–52). San Diego, CA: Academic Press.

Roswarski, T. E., & Proctor, R. W. (1996). Multiple spatial codesand temporal overlap in choice-reaction tasks. PsychologicalResearch, 59, 196–211.

Roswarski, T. E., & Proctor, R. W. (2000). Auditory stimulus-response compatibility: Is there a contribution of stimulus-handcorrespondence? Psychological Research, 63, 148–158.

Rugg, M. D., & Coles, M. G. H. (Eds.). (1995). Electrophysiologyof mind: Event-related brain potentials and cognition. Oxford,England: Oxford University Press.

Sanders, A. F. (1975). The foreperiod effect revisited. QuarterlyJournal of Experimental Psychology, 27, 591–598.

Sanders, A. F. (1998). Elements of human performance. Mahwah,NJ: Erlbaum.

Sanders, A. F., & Wertheim, A. H. (1973). The relation betweenphysical stimulus properties and the effect of foreperiod durationon reaction time. Quarterly Journal of Experimental Psychol-ogy, 25, 201–206.

Schweickert, R. (1983). Latent network theory: Scheduling ofprocesses in sentence verification and the Stroop effect. Journalof Experimental Psychology: Learning, Memory, and Cognition,9, 353–383.

Seibel, R. (1963). Discrimination reaction time for a 1,023-alternativetask. Journal of Experimental Psychology, 66, 215–226.

Shaffer, L. H. (1965). Choice reaction with variable S-R mapping.Journal of Experimental Psychology, 70, 284–288.

Smith, E. E. (1968). Choice reaction time: An analysis of the majortheoretical positions. Psychological Bulletin, 69, 77–110.

Soetens, E. (1998). Localizing sequential effects in serial choicereaction time with the information reduction procedure. Journalof Experimental Psychology: Human Perception and Perfor-mance, 24, 547–568.

Sternberg, S. (1969). The discovery of processing stages: Exten-sions of Donders’ method. In W. G. Koster (Ed.), Attention andperformance II. Acta Psychologica, 30, 276–315.

Sternberg, S. (1998). Discovering mental processing stages: Themethod of additive factors. In D. Scarborough & S. Sternberg(Eds.), An invitation to cognitive science: Methods, models, andconceptual issues (2nd ed., Vol. 4, pp. 703–863). Cambridge,MA: MIT Press.

Stoffer, T. H., & Umiltà, C. (1997). Spatial stimulus coding andthe focus of attention in S-R compatibility and the Simon ef-fect. In B. Hommel & W. Prinz (Eds.), Theoretical issues instimulus-response compatibility (pp. 181–208). Amsterdam:North-Holland.

Stroop, J. R. (1992). Studies of interference in serial verbal reac-tions. Journal of Experimental Psychology: General, 121,15–23. (Original work published 1935)

Tagliabue, M., Zorzi, M., Umiltà, C., & Bassignani, F. (2000). Therole of LTM links and STM links in the Simon effect. Journal ofExperimental Psychology: Human Perception and Performance,26, 648–670.

Teichner, W., & Krebs, M. J. (1974). Laws of visual choice reactiontime. Psychological Review, 81, 75–98.

Telford, C. W. (1931). The refractory phase of voluntary andassociative responses. Journal of Experimental Psychology, 14,1–36.

Umiltà, C., & Nicoletti, R. (1990). Spatial stimulus-response com-patibility. In R. W. Proctor & T. G. Reeve (Eds.), Stimulus-response compatibility: An integrated perspective (pp. 89–116).Amsterdam: North-Holland.

Van Zandt, T., Colonius, H., & Proctor, R. W. (2000). A comparisonof two response time models applied to perceptual matching.Psychonomic Bulletin & Review, 7, 208–256.

Vickrey, C., & Neuringer, A. (2000). Pigeon reaction time, Hick’slaw, and intelligence. Psychonomic Bulletin & Review, 7,284–291.

Vu, K.-P. L., & Proctor, R. W. (2001). Determinants of right-left andtop-bottom prevalence for two-dimensional spatial compatibil-ity. Journal of Experimental Psychology: Human Perception andPerformance, 27, 813–828.

Vu, K.-P. L., & Proctor, R. W. (2002). Mixing compatible andincompatible mappings: Elimination, reduction, and enhance-ment of spatial compatibility effects. Manuscript submitted forpublication.

Wang, H., & Proctor, R. W. (1996). Stimulus-response compatibilityas a function of stimulus code and response modality. Journal ofExperimental Psychology: Human Perception and Performance,22, 1201–1217.

Weeks, D. J., & Proctor, R. W. (1990). Salient-features coding inthe translation between orthogonal stimulus and response di-mensions. Journal of Experimental Psychology: General, 119,355–366.

Weeks, D. J., Proctor, R. W., & Beyak, B. (1995). Stimulus-responsecompatibility for vertically oriented stimuli and horizontally ori-ented responses: Evidence for spatial coding. Quarterly Journalof Experimental Psychology, 48, 367–383.

Welford, A. T. ( 1987). Comment on “Exceptions to Hick’s law:Explorations with a response duration measure” (Longstreth,El-Zahhar, & Alcorn, 1985). Journal of Experimental Psychol-ogy: General, 116, 312–314.

Wickelgren, W. A. (1977). Speed-accuracy tradeoff and informationprocessing dynamics. Acta Psychologica, 41, 67–85.

Woodworth, R. S. (1938). Experimental psychology. New York:Holt.

Wundt, W. (1883). Über psychologische Methoden [Over psycho-logical methods]. Philosophische Studien, 10, 1–38.

CHAPTER 12

Motor Control

HERBERT HEUER

317

THE PROBLEM OF MOTOR CONTROL 317An Outline of the Problem 317An Outline of Possible Solutions 319Indeterminateness of the Solutions 320

MOTOR PREPARATION 322The Anticipatory Nature of Motor Preparation 323Motor-Control Structures 324The Advance Specification of Movement

Characteristics 326THE USE OF SENSORY INFORMATION 327

Target Information 329Feedback Information 333

Sensory Information for Motor Control andPerception 334

MOTOR COORDINATION 335Task Constraints and Structural Constraints 335Basic Structural Constraints on Coordination 337Levels of Coupling 340

FLEXIBILITY OF MOTOR CONTROL 342Adapting and Adjusting to New Visuo-Motor

Transformations 342Adjusting and Adapting to External Forces 344

MOVING ON 346REFERENCES 346

Motor control is a cross-disciplinary field of research inwhich the boundaries between established academic disci-plines like psychology, physiology, neurology, engineering,and physical education are blurred. Within psychology,motor behavior tended to be a rather marginal topic for vari-ous reasons. When psychology is conceived as a science ofthe mind, movement is more or less beyond its scope. Lessobviously, even when psychology is conceived as a scienceof behavior, issues of motor control do not become focal; forexample, behaviorism was more concerned with “what isdone” questions than with “how is it done” questions. Finally,although the first well-known psychology paper on motorcontrol appeared at the end of the nineteenth century(Woodworth, 1899), and although James (1890, 1950) de-voted a chapter to “The Production of Movement,” touchingon the topic in several other chapters, the founding fathers ofpsychology did not stamp motor control as an essential ingre-dient of the emerging academic discipline.

The field of motor control gains in importance as soon asone envisages that the human mind and brain may haveevolved primarily to support action, not to contemplate theworld. Then the question of how goals can be reached be-comes critically important. This question alludes to problemsof control, and motor control deals with particular goals thatcan be reached by moving one’s limbs.

In this chapter I first introduce the core problem of motorcontrol and discuss different ways that it can be solved.Basically, there are two such ways: open-loop and closed-loop control. Open-loop processes are initiated before amovement is actually executed, so they are described underthe heading of motor preparation. The next section then dealswith closed-loop processes, the exploitation of sensory feed-back from an ongoing movement in the service of motor con-trol, but also with other uses of sensory information. After thediscussion of these rather fundamental issues, the perspectiveis enlarged somewhat. Many motor skills require coordinatedmovements of different limbs, which opens the topic ofmotor coordination. Finally, I shall address the flexibilityof motor control which enables us to operate various toolsand machines and to handle objects of various masses.

THE PROBLEM OF MOTOR CONTROL

An Outline of the Problem

Movements result from an interplay of passive and activeforces. Passive forces are due to our own movements as well asto environmental factors like gravity. For example, in theswing phase of the walking cycle the thigh is rotated forward;

318 Motor Control

F E

M

F1 F2

(b)

(a)

Figure 12.1 (a) A joint with two opposing muscles, flexor (F) and extensor(E); (b) mechanical analogue of two damped springs acting on a single mass(M); the mass is stationary when the two opposing forces F1 and F2 canceleach other. Reprinted with permission.

initially the knee is flexed, followed by extension. This for-ward rotation of the shank results largely from passive forcesof different origins. The deceleration of the knee extension,however, is largely a result of active muscular forces, with onlya small contribution of passive ones (Winter & Robertson,1978). Thus, with the exception of a few very simple tasks, theproduction of movement requires not only the generation ofappropriate active forces, but in addition passive forces have tobe taken into account.

Figure 12.1 illustrates a joint with two opposing muscles,a kind of minimal movement device. Muscles are designatedas agonist and antagonist with respect to their function in aparticular movement. For example, when the movement is aflexion of the joint, the flexor is the agonist and the extensoris the antagonist; for an extension, the functional roles offlexors and extensors are reversed. Of course, Figure 12.1 isextremely simplified, both with respect to the mechanicalcharacteristics and with respect to the number of muscles act-ing on the joint.

Muscles are complicated force generators. They contractwhen they are activated via the motor nerves. Each axon of amotor nerve innervates a smaller or larger bundle of musclefibers; the axon together with its muscle fibers is called amotor unit. The activation can be recorded. Needle elec-trodes, which are inserted in the muscle tissue, allow one torecord from single motor units, while surface electrodes pickup averaged and filtered electrical activity of motor unitswithin a certain area below the electrodes. For isometric con-

tractions, there is a systematic relation between electromyo-graphically recorded muscle activity (EMG) and force. Inparticular, the relation between the integrated EMG signaland force is linear (Lippold, 1952). However, for movementsfor which phasic bursts of muscle activity are typical (atleast when the movements are rapid), the relation is morecomplex.

Complications arise, first, from the temporal relations be-tween bursts of muscle activity and forces, which can befairly variable. In general, forces develop only with a delaywhen a muscle is activated, and after the end of the burstthere is a gradual decay. Complications arise also fromfatigue-induced changes, with fatigue being developed inthe course of repeated or prolonged activity. In addition, fora given activation level, muscle force depends on the lengthof the muscle and on the rate of its contraction. In particular,the length-tension relation of muscle is important for modelsof motor control: Muscle force increases with increasingmuscle length, and the slope becomes steeper the strongerthe activation of the muscle is (e.g., Rack & Westbury,1969). Although the length-tension relation is not really lin-ear, a linear approximation is useful, at least for certainranges of muscle length. Thus, one can think of a muscleas being mechanically similar to a damped spring (cf. Fig-ure 12.1).

A muscle can actively contract, but not stretch. (A rubberband would perhaps be a better analogue than a spring.)Therefore at least two opposing muscles are needed for asimple joint. From Figure 12.1 it is apparent that, as the onemuscle contracts, the other one will be stretched. This impliesthat, with given activations of the opposing muscles, theforce of the contracting muscles declines while that of thestretched muscles increases. At a certain joint angle, and at acertain relation between the lengths of the opposing muscles,the forces developed by them will be equal, but in oppositedirections, and thus cancel each other. The net force is zero,and the joint position at which this is the case is called theequilibrium position. There is considerable evidence thatequilibrium positions are important for motor control(cf. Kelso & Holt, 1980; Polit & Bizzi, 1979). In the simplestversion of a mass-spring model, movements come about sim-ply by the specification of a new equilibrium position (e.g.,Cooke, 1980), but experiments have revealed that the equi-librium position shifts continuously and not stepwise (Bizzi,Accornero, Chapple, & Hogan, 1984).

Movement results from the net force of opposing muscles(and, of course, from passive forces). Thus, at first glancethere seems not to be much sense in cocontractions, in whichopposing muscles are active simultaneously. Nevertheless,cocontractions can be observed in particular early during

The Problem of Motor Control 319

�1

�3

�2

(x1, x2)

Figure 12.2 A three-jointed arm with joint angles �1, �2, and �3, which areassociated with spatial position (x1, x2) of the end-effector.

Figure 12.3 Possible solutions to a control problem. (a) Open-loopsolution with an internal model Tˆ –1 of the (inverse) transformation;(b) closed-loop solution, with C as controller; (c) combination of open-loopand closed-loop solution.

practice (e.g., Metz, 1970) and in tasks requiring high preci-sion. Even when no net forces result from cocontractions,they modulate the mechanical characteristics of the joint likefriction.

Saying that joint movement results from the net force ofopposing muscles (in addition to passive forces) is not thewhole story. More precisely, joint rotation results from thetorque, which again is related to the net force in a fairly com-plicated way, with the relation being dependent on the jointangle. Even with the movement of the joint, the sequence oftransformations from muscle activation to movement has notyet reached its end, because in general the goals for ourmovements are not defined in terms of joint angles.

Figure 12.2 illustrates a three-jointed arm with the end-effector pointing to a target. The goals of many movementsare defined in terms of reaching for some spatial target; forother movements, as in catching a ball, there are temporal tar-gets in addition; for still other movements, as in writing,goals are defined in terms of movement traces (or paths).From Figure 12.2 it is apparent that a particular configurationof joint angles is associated with a particular spatial positionof the end-effector.

Thus far I have sketched the transformation of muscle ac-tivation to the spatial position of an end-effector like the tipof the index finger. The purpose was to give some impressionof the complexity of this transformation without going intotoo much detail. Sometimes different components of thetransformation are discussed separately, in particular thekinematic transformation (from joint angles to end-effectorpositions) and the dynamic transformation (from torques tomovements of the joints). As a more general term, I shall usemotor transformation to refer to the total transformation orsome part of it.

Given the complexity and the time-varying characteristicsof the motor transformation, one may wonder that humans—at least after the first few months of their life—are able to

produce purposeful movements at all, and not only random-appearing ones. This requires that humans be able to deter-mine the pattern of muscular activity that is required toproduce a particular movement of a particular end-effector.The very fact that humans can produce purposeful move-ments indicates that nature has solved this core problem ofmotor control; what remains for the movement scientist is togain an understanding of what the solution is.

An Outline of Possible Solutions

The core problem of motor control can be stated in a verysimple and general way. Let T be a transformation of an inputsignal x into an output signal y. For example, y shall be aparticular time-varying position of an end-effector, and x avector that captures time-varying muscle activity. Then thegeneral problem of control, and that of motor control in par-ticular, is to determine an input signal x such that the outputsignal y becomes identical to the desired output signal y*.The problem is solved when the inverse of the transformationT can be determined, such that T –1T = 1. Thus, control re-quires the inversion of a transformation, and there are twofundamentally different ways to achieve this (see Jordan,1996, for a detailed discussion).

Figure 12.3a illustrates an open-loop solution which re-quires an internal model T^ –1 of the transformation, or, moreprecisely, of its inverse. There are different ways of imple-menting such a model formally (e.g., Jordan, 1996). Of

320 Motor Control

course, such an internal model is not necessarily a kind of en-tity, located in some part of the brain, but it can result fromthe activity of a network that is distributed widely across bothcentral and more peripheral levels of the motor system (e.g.,Kalveram, 1991).

Figure 12.3b illustrates a closed-loop solution for whichno internal model T^ –1 is required. Instead, the inversion ofthe transformation results from the structure of the loop.(This is shown formally by Jordan, 1996.) Intuitively this be-comes clear from the following consideration. A closed-loopsystem can reduce the deviation between the output y and thedesired output y*. To the extent that this is successful, y andy* become similar. This then implies, because y = T(x), thatx approximates T –1(y*).

For some years, open-loop and closed-loop models ofmotor control were contrasted (cf. Stelmach, 1982). How-ever, by now it is clear that nature combines both types ofsolution, roughly in a way illustrated in Figure 12.3c. Thiscombination maintains the advantages of both types of solu-tion and avoids the disadvantages of each of them. In addi-tion, the combination exhibits some characteristics thatmatch characteristics of human movements (Cruse, Dean,Heuer, & Schmidt, 1990).

The disadvantage of an open-loop solution is its limitedprecision. The motor transformation is complex, and it hastime-varying characteristics. When we use tools or operatemachines, there are additional transformations that must betaken into account, like the transformation of a steering-wheelrotation into a change of the direction in which a vehicle isheading. Thus, internal models of inverse transformations canonly be approximations. The disadvantage of a closed-loopsystem is that it involves time delays and can become insta-ble, in particular when the gain is high. On the other hand, ahigh gain is desirable to improve accuracy. When both sys-tems are combined, open-loop control will serve to approxi-mate the desired output; closed-loop control is suited toreducing the remaining deviation even when the gain is rela-tively small, which serves to avoid instabilities.

There are two different types of procedure to determinewhether a control system is closed-loop or open-loop. Thefirst is to cut the potential feedback loop, and the second is todistort the potential feedback signal. Both manipulationsshould have essentially no effect when the control systemis open-loop, but strong effects when the control system isclosed-loop; with eliminated feedback, the closed-loop sys-tem should produce no change of the output signal or onlyrandom changes, and with distorted feedback the outputshould be distorted. Human movements are often little af-fected by elimination of feedback, but strongly affected by itsdistortion. Such results do not give a clear answer with

respect to the dichotomy of open-loop versus closed-loopcontrol, but they conform to expectations based on the com-bined control modes (Cruse et al., 1990).

Indeterminateness of the Solutions

Typically movements are not fully determined by their goals.An example is reaching, with the goal being defined in termsof a spatial target position. Thus, only the endpoint of themovement is specified by the goal, but not its time-course. Inspite of this indeterminateness a solution is reached, whichtakes additional task constraints as well as organismic con-straints into account.

Perhaps the most extensively studied task constraint is thesize of the spatial target, which affects movement durationand the shape of velocity-time curves (e.g., MacKenzie,Marteniuk, Dugas, Liske, & Eickmeier, 1987). Basically, forsmaller targets humans choose to produce slower move-ments. The relation of movement time not only to targetwidth, but also to the distance of the target from the start po-sition, is of a particular kind known as Fitts’ law. The early1950s, when Fitts (1954) first described the relation, saw therise of information theory in psychology. Thus, the relationwas formulated in terms of information measures, and thetradition has left it in that form. Fitts’ law states that move-ment time is a linear function of the index of difficulty, whichis defined as log2(2A�W), A being the movement amplitudeand W the width of the target.

Fitts’ law describes a particular kind of speed-accuracytrade-off: Faster movements have a larger scatter of theirend-positions than slower movements, so when a small scat-ter is required because the target is small, slower movementshave to be chosen. The law is astonishingly robust (cf. Keele,1986), and it has given rise to various theoretical accounts(Crossman & Goodeve, 1963/1983; Fitts, 1954; Meyer,Abrams, Kornblum, Wright, & Smith, 1988), but also to al-ternative formulations (cf. Plamondon & Alimi, 1997) and tocontrasting observations (e.g., Schmidt, Zelaznik, Hawkins,Frank, & Quinn, 1979), in particular for situations that re-quire a certain movement duration, rather than reading aspatial target of a particular width. (Wright & Meyer, 1983;Zelaznik, Mone, McCabe, & Thaman, 1988).

Although they have received much less attention, othertask constraints than target size affect the chosen movementtrajectory. For example, it makes a difference whether thespatial target has to be hit or whether an object in the sameposition has to be grasped, and in the latter case it makes adifference whether the object is a tennis ball or a light bulb.The movement to the light bulb takes more time than themovement to the tennis ball; in particular, the deceleration of

The Problem of Motor Control 321

Figure 12.4 (a) A complex movement pattern produced under the instruc-tion to maintain a constant velocity; (b) circles produced under the instruc-tion of a constant velocity (left) and a time-varying velocity (right). Theseexamples are taken from Derwort (1938). Recordings were made with a lightplaced on the index finger. The shutter of a camera was opened about60 times per second; distance between dots thus represents distance coveredin 1�60 s, smaller distances indicating smaller, and larger distances higher,velocity.

the movement toward the bulb is more gradual and extendedin time (Marteniuk, MacKenzie, Jeannerod, Athènes, &Dugas, 1987). Another task constraint has been reported re-cently: The time it takes to move a mug to the mouth dependsin a particular way on the diameter of the mug and the dis-tance from the level of water to the edge (Latash & Jaric, inpress). Such task constraints are at least to some degree re-flected by our everyday experience.

A second type of constraints, which are taken into accountwhen movement trajectories are indeterminate, is of a moreorganismic nature and related to the costs of movements.Although the general notion of cost minimization—as far asthis is possible with the given task constraints—has a highdegree of plausibility, it poses more of a problem than asolution. There are many different kinds of costs that can po-tentially be minimized. For example, Nelson (1983) analyzedthe consequences of minimizing five different kinds of costsfor the trajectories of movements aimed at a target. Othercriteria have been added (e.g., Cruse, 1986; Cruse & Brüwer,1987; Rosenbaum, Slotta, Vaughan, & Plamondon, 1991;Rosenbaum, Vaughan, Barnes, & Jorgensen, 1992; Uno,Kawato, & Suzuki, 1989), and perhaps any list will be in-complete.

A fairly general principle seems to be that movement tra-jectories are selected by the criterion of smoothness. Al-though in principle smoothness can be defined in differentways, one of the possible criteria is minimization of jerk, thatis, minimization of the integral of the squared third derivativeof end-effector position with respect to time (Flash & Hogan,1985). The principle can be extended and used to model com-plex movement patterns, as in handwriting (cf. Teulings,1996). In addition, for drawing-like movements, it producesa particular relation between curvature and tangential veloc-ity, which is known as the two-thirds power law (Viviani &Flash, 1995). Basically, with a larger radius of curvature,velocity tends to be higher than with a smaller radius of cur-vature even when the instruction is to maintain a constant ve-locity (Figure 12.4). The dependency of velocity on curvatureis particularly conspicuous in drawing ellipses for whichthe radius of curvature varies continuously. Although the re-verse relation has received less attention, variations of veloc-ity do also induce variations of curvature; for example, whenone attempts to draw circles with a pattern of smaller-higher-smaller-higher velocity within each cycle, the result is likelyto be ellipses (Derwort, 1938).

Indeterminateness does exist even when the goal of amovement specifies a trajectory of the end-effector in everydetail. Of course, in such cases the movement trajectory isnot indeterminate, but the input to the motor transformationis. The origin of the indeterminateness is apparent from

Figure 12.2, where the target position is specified in terms oftwo spatial dimensions, but it can be reached with differentconfigurations of three joints. More generally, the output ofthe motor transformation has a lower dimensionality than theinput, so that the inversion of the motor transformation has nounique solution. The problem of how to deal with the manydimensions of the input is often called the degrees-of-freedom problem. A consequence is motor equivalence: Thesame movement can be performed in many different ways.

Again, cost minimization can be considered as a way toreach a unique solution (cf. Cruse, 1986; Cruse & Brüwer,1987; Rosenbaum et al., 1991). Another possibility is thefreezing of degrees of freedom. For example, in handwritingadults mainly use the wrist and the fingers, and hardly or notat all the elbow and the shoulder joints. When one observespreschoolers at their first attempts to write (which might notbe the appropriate term for the result, but perhaps for the in-tention), one can notice that the wrist and fingers are largelyimmobilized, and that mainly the more proximal joints,which are closer to the trunk, are used (Blöte & Dijkstra,1989). This can also be observed when adult right-handerswrite with their left hand (Newell & van Emmerik, 1989).Finally, the high dimensionality of an input vector can bereduced to a small number of degrees of freedom by way ofintroducing covariations. A somewhat trivial example againcan be seen in handwriting: With a normal tripod grip, thumb,index finger, and middle finger are mechanically coupled(because of holding the pen) and can no longer be movedindependently.

Motor equivalence implies not only the existence ofcriteria for selecting one of the many options, but also thatdifferent options can be chosen in case that it is desirable ornecessary. For example, when one asks people to tap withtheir index finger as rapidly as possible, and to do so as long

322 Motor Control

Figure 12.5 Example recording of a rapid finger flexion.

as they can, several of them will gradually replace move-ments of the finger with movements of the wrist. Less inci-dentally, Lippold, Redfearn, and Vuco (1960) describe whatthey call “migration of activity” from one muscle to otherones during prolonged activity that induces muscular fatigue.More generally, the many-to-few mapping of the motor trans-formation leaves the option to select different subsets fromthe many input dimensions when some of them are function-ally impaired, be it a fatigued muscle or an immobilized joint.

MOTOR PREPARATION

The initiation of a movement is a gradual and continuousprocess. In Figure 12.5 an example recording of a rapidindex-finger flexion of about 20° amplitude is shown, as arein particular the position-time curve, the velocity-time curve,the acceleration-time curve, and the EMG of a finger flexor(agonist) and an extensor (antagonist). Faced with suchrecordings, it becomes somewhat difficult to answer thequestion of when the movement starts. Typically the start of amovement is defined in terms of a threshold for one of thekinematic signals. From Figure 12.5 it is apparent that defin-itions based on the acceleration signal generally lead toearlier initiation times than definitions based on the positionsignal: There can be a sizeable acceleration while positionhas hardly changed. Thus, any definition of the start of amovement is to some degree arbitrary.

Muscle activity can be observed in advance of changes ofkinematic signals, and the definition of the start of a movementcan also be based on EMG traces. In many instances the agonistburst is over before a change of position can be seen. Thus, it isnot too remarkable that the agonist burst is hardly or not at allaffected when the overt movement is unexpectedly blocked.More remarkable is that the later bursts, which normallyserve to decelerate the limb and to stabilize the end-position,still occur, although they serve no obvious purpose any more(Wadman, Denier van der Gon, Geuze, & Mol, 1979).

The overt movement is preceded not only by muscle activ-ity, but also by various kinds of preparatory processes whichcan be evidenced at different levels of the motor system (seeBrunia, 1999, and Brunia, Haagh, & Scheirs, 1985, foroverviews of psychophysiological findings). For example, inthe electroencephalogram, movement-related activity can beseen when the start of the movement is used as a trigger for av-eraging. Even such simple voluntary movements as key-presses are preceded by a slowly increasing negativity thatstarts in the order of 1 s before the overt movement. This readi-ness potential or Bereitschaftspotential was first described byKornhuber and Deecke (1965). Initially it is symmetrical, butin the last 100 or 200 ms it becomes asymmetrical, being

stronger over the hemisphere contralateral to the respondinghand. This kind of asymmetry can also be observed in reac-tion-time tasks. Called the lateralized readiness potential, ithas become an important tool in information-processing re-search (see chapter by Proctor & Vu in this volume).

Motor Preparation 323

Figure 12.6 (a) Response panel used by Rosenbaum et al. (1992). The barwith pointer had to be grasped and to be placed on one of the eight targetswith the pointer toward the LED (black dots). (b) Relative frequency ofgrasping the bar with the thumb toward the pointer as a function of thefinal orientation (targets 1–8), shown separately for four different initialorientations.

The Anticipatory Nature of Motor Preparation

The psychophysiological data indicate the existence of motorpreparation, but they are more or less silent to the question ofwhat goes on in functional terms. What they tell, of course, isthat at least to some degree preparatory processes are specificfor the forthcoming movement in that the data reflect some ofits characteristics, like the hand used. Näätänen and Merisalo(1977) suggested that the essence of motor preparation is thateverything is done in advance of the overt response that canbe done, which amounts to activating the response up to alevel close to the motor action limit. This characterization ofmotor preparation may be appropriate for simple movementslike keypresses, but it falls short of capturing essential char-acteristics of motor preparation preceding a more complexmovement.

Preparatory activities in general can anticipate the futureto varying degrees. For example, in preparing for a vacation,one might book a hotel in advance for only the first night, orone might book hotels in different places for several nights tocome. Activating a response close to the action limit meanspreparing only for movement initiation (like booking a hotelfor the first night). However, motor preparation is also con-cerned with the future of the response (like booking hotels forseveral nights to come). There are at least three kinds of evi-dence for this.

The first kind of evidence is from reaction-time experi-ments. When the task is to perform a sequence of simplemovements, simple reaction time increases with the length ofthe sequence. The seminal study was by Henry and Rogers(1960), who found increasing reaction times for (a) lifting afinger from a key, (b) lifting the finger from the key andgrasping a tennis ball at a certain distance, and (c) lifting thefinger from the key, touching the ball with the back of thehand, pressing another key, and hitting a second tennis ball.More systematic explorations of the sequence-length ef-fect have been reported by Sternberg and coworkers (seeMonsell, 1986, for an overview). With more homogeneouselements like keypresses, letter names, or words with a cer-tain number of syllables, reaction time increases linearly withthe number of sequence elements. At some length of about6–12 elements, the increase of reaction time levels off, earlierfor longer elements (like trisyllabic words) and later forshorter elements (like monosyllabic words).

The second kind of evidence is from studies of anticipatorypostural adjustments (see Massion, 1992, for review). When aforthcoming movement threatens balance, the voluntary ac-tion is preceded by the appropriate postural adjustments. Forexample, Cordo and Nashner (1982) observed EMG activityof postural muscles in the leg of their standing subjects whichpreceded by about 40 ms the activity of arm muscles involved

in the task of pulling a hand-held lever in response to an audi-tory signal. In a control condition with a passive support, thepreparatory postural activity was absent, and arm-muscle ac-tivity had a shorter latency. Thus, anticipatory postural adjust-ments are not only specific with respect to the forthcomingvoluntary movement (e.g., Zattara & Bouisset, 1986), but alsowith respect to context characteristics.

The third kind of evidence, finally, shows that earlier partsof a motor pattern are adapted to later parts. Evidence forthis can be found in many skills (cf. Rosenbaum & Krist,1996), but I shall focus here on a particularly basic kind ofobservation, the effect of end-state comfort (Rosenbaum &Jorgensen, 1992; Rosenbaum et al., 1992). Figure 12.6a illus-trates the task of Rosenbaum et al. (1992, Exp. 1). The stand-ing subject had to grasp a bar with a pointer, which haddifferent initial orientations, the pointer pointing upward,downward, to the left or to the right. With the pointer upwardor to the left, it is quite comfortable to grasp the bar with thethumb toward the pointer, but with the other two initial ori-entations this is less comfortable. Under speed instructions

324 Motor Control

the subjects had not only to grasp the bar, but also to place itin one of eight target positions with the pointer toward theLED that signaled the target position in each trial; thus, therewere differences in final orientation. For targets 6–8 and 1–2,holding the bar in the final orientation with the thumb towardthe pointer is more or less comfortable, but for targets 3–5,holding the bar with the thumb away from the pointer is morecomfortable. In Figure 12.6b the relative frequency of grasp-ing the bar in its initial position with the thumb toward thepointer is shown. These data reveal not only an effect of theinitial orientation of the bar, but also a clear effect of the finalorientation. Thus, the effect of end-state comfort can be evi-denced at the very start of the action, and it clearly indicatesthat motor preparation embraces anticipation.

The anticipatory nature of motor preparation implies thatthere is some kind of representation of the forthcomingmovement before it begins. The existence of such a represen-tation also implies that open-loop processes of motor controlare of a particular nature in that they are predictive. In fact,the answer to the question of what goes on during motorpreparation in functional terms may be largely that this kindof internal representation of the forthcoming movement is setup, which then allows for a more or less autonomous control.

Motor-Control Structures

There are different ways to conceptualize autonomousprocesses of motor control. In psychology it had beencommon to designate the anticipatory representation of aforthcoming movement as a motor program (and the processof setting it up as programming). However, this term has be-come associated with a particular conceptualization. There-fore, as a broader and more neutral term, Cruse et al. (1990)have suggested motor-control structures. There seem to bebasically two different ways of modeling them, either interms of prototypical functions or in terms of generativestructures (Heuer, 1991).

Prototypical Functions

Movements vary qualitatively as well as quantitatively. Oneof the attempts to capture this basic observation is the notionof a generalized motor program, most explicitly introducedby Schmidt (1975). A generalized motor program is thoughtto control a set of movements that have certain characteristicsin common. The specifics of each particular movement arethought to be determined by the program’s parameters. Thus,for a certain type of movement there should be invariant char-acteristics, which represent the signature of the program, andvariable characteristics, which reflect the variable settings of

its parameters. Of course, such a concept requires that theinvariant characteristics of movements of a certain type beidentified.

The theoretical problem of identifying invariant character-istics met with observations of an invariance of relative tim-ing in different motor skills (see Gentner, 1987, for a review),which led Schmidt (1980, 1985) to propose that the relativetiming is an invariant feature of movements that are con-trolled by a single generalized motor program. In addition,relative force was hypothesized to be a second invariant char-acteristic. With these assumptions, a generalized motor pro-gram can be described by way of a prototypical force-timefunction �(�), which can be scaled in time by a rate parame-ter and in amplitude by a force parameter.

The notion of a prototypical force-time function, whichcan be scaled in time as well as in amplitude, is reminiscentof the way we use coordinate systems to represent force-timecurves. Thus one might suspect that the concept is relatedmore to how we plot force as a function of time than to howthe brain controls movement. Nevertheless, the notion isnot biologically implausible. One can think of a spatiallyorganized representation that is read at a certain rate andthus transformed into a temporally organized movement(cf. Lashley, 1951). The speed of reading would correspondto the rate parameter. Similarly, as the read signal is chan-neled to the muscles, it could be amplified to variable degrees(cf. von Holst, 1939). Thus, in principle, the notion of proto-typical functions implies a certain degree of independence oftemporal control and force control.

The most detailed application of prototypical force-timefunctions has been in models of the speed-accuracy trade-off in rapid aimed movements. These so-called impulse-variability models account for the trade-off in terms ofnoise in the motor system (Meyer, Smith, & Wright, 1982;Schmidt, Sherwood, Zelaznik, & Leikind, 1985; Schmidtet al., 1979). However, it is not really necessary that proto-typical curves specify forces; instead, they can also bethought of as specifying kinematic characteristics (e.g.,Heuer, Schmidt, & Ghodsian, 1995; Kalveram, 1991). Infact, formal models of the autonomous processes of motorcontrol are generally somewhat diverse or even indetermi-nate with respect to their output variables.

The motor transformation involves a number of differentvariables, and in principle any of these can be taken as outputvariable for models of motor-control structures. Ultimately,of course, muscles must be activated. In fact, the concept of amotor program has often been associated with a prestructuredsequence of muscle commands (Keele, 1968). At the otherextreme, motor-control structures can be modeled with thetrajectory of the end-effector as the output. In the first case,

Motor Preparation 325

the inversion of the motor transformation is assumed to be anintegrated component of a motor-control structure. In the sec-ond case, it is left to additional and separate processes. Al-though the choice may be somewhat arbitrary, it implies anassumption about whether the internal model of the inversemotor transformation is specific for a particular type ofmovement governed by a particular motor-control structure,or whether it is generalized and thus applicable to differenttypes of movement.

There are some considerations and data that favor themodeling of motor-control structures with end-effector kine-matics as output. One consideration starts with the observa-tion that both perception and action are externalized. Forexample, we do not see the image on the retina, but objectsand their locations in the world. Similarly, awareness of ourown movements is typically not in terms of muscular con-tractions and joint angles. Visual distances and movementamplitudes in the external world are commensurate, whereasproximal visual stimuli and patterns of muscular activity arenot (cf. Prinz, 1992). Thus, to be compatible with how weperceive the world around us, movement should be repre-sented in terms of world coordinates.

Another consideration starts with the assumption that thevariables used in motor preparation or planning should revealthemselves by the possibility of describing them concisely aswell as by their consistency. For example, for pointing in atwo-dimensional plane as in Figure 12.2, the movement pathsapproximate straight lines, whereas the relations betweenjoint-angles can be fairly complex. More specifically, plottingthe y coordinates of the end-effector as a function of the xcoordinates results in straight lines at least approximately,whereas plotting the elbow angle as a function of the shoul-der angle results in strongly curved lines. This suggests thatmotor-control structures deal with the trajectory of the end-effector, and that the time-courses of joint angles are a conse-quence thereof (cf. Hollerbach & Atkeson, 1987). Similarly,kinematic characteristics of single-joint movements arehighly similar for movements with and against gravity,whereas the patterns of muscular activity are grossly different(Virjii-Babul, Cooke, & Brown, 1994).

No matter for which kind of variable prototypical func-tions are defined, the notion is intimately related to the invari-ance of relative timing. The invariance is never really perfect,but often. It can be taken as a reasonable approximation.However, there are also clear deviations from invariance. Forexample, when the target size is reduced or accuracy ratherthan speed is emphasized, the relative duration of the deceler-ation phase of aimed movements tends to increase (Fisk &Goodale, 1989; MacKenzie et al., 1987). Moreover, theconcept of a prototypical function takes a particular relative

timing as a mandatory characteristic of a certain type ofmovement which cannot easily be changed; however, whenafter some practice in a particular temporal pattern the relativetiming is changed, humans do not encounter particular diffi-culties (Heuer & Schmidt, 1988). Thus, prototypical func-tions do not represent a valid type of model for motor-controlstructures in general, but nevertheless they can capture im-portant characteristics of some types of movement.

Generative Structures

Whereas a conceptualization of motor-control structures interms of prototypical functions posits stored trajectories, con-ceptualizations in terms of generative structures posit net-works that generate the trajectories. An example is a modelby Saltzman and Kelso (1987) that belongs to a class theycalled the “task-dynamic approach.” For an aimed movement,Saltzman and Kelso defined a reach axis that runs through thetarget and the current position of the end-effector as well as anaxis orthogonal to it. These axes define an abstract task spacein which the end-effector is represented by a “task mass.” Thetarget position is located in the origin of the task space and isassumed to have the characteristics of a point attractor. Thus,wherever the task mass is in task space, it will move towardthe target governed by a set of simple equations of motion; forthe reach axis x it is mT x + bT x + kTx = 0, with the index Tdesignating parameters of the task space.

The task-dynamic approach goes beyond advance specifi-cations of movements in task space. For example, jointmovements are derived by way of coordinate transforms.However, for the present purpose only the highest level of thescheme is important. At first glance there does not seem to bemuch difference between describing a motor-control struc-ture in terms of a differential equation that governs a genera-tive structure or in terms of a solution of such an equation thatcould be stored as a prototypical function. However, thereare differences. First, the parameterizations are different.Whereas the prototypical function has a rate and an ampli-tude parameter, the particular generative structure at handhas abstract mass, mT, friction, bT, and stiffness, kT, parame-ters. Variation of these parameters, for example, does notnecessarily result in relative-timing invariance. Second, andperhaps more important, the generative structure is lesssusceptible to the effects of transient perturbations. Itimplements a movement characteristic called equifinality:Movements tend to reach their target even when they aretransiently perturbed (Kelso & Holt, 1980; Polit & Bizzi,1979; Schmidt & McGown, 1980).

Although the model of Saltzman and Kelso (1987) seemsto be more mathematically than physiologically inspired, this

326 Motor Control

Figure 12.8 Change of the direction of the population vector as a functionof the time since presentation of an imperative stimulus (after Georgopouloset al., 1989).

is different with the VITE model of Bullock and Grossberg(1988). (VITE stands for vector-integration-to-endpoint.)The formal structure of an element of the model is illustratedin Figure 12.7. The variable P is an internal representation ofthe position of an effector, and T represents a target position.The variable V represents the (delayed) difference, and G theGo signal. In principle, the structure of Figure 12.7 is thoughtto be multiplied for different muscles that are involved in avoluntary movement, with V ≥ 0 for each particular muscle.

Without going into mathematical details, it is worth notingthat the difference V in the case of aimed movements is againgoverned by a second-order differential equation (providedthat G is a constant). In spite of this similarity, there are sev-eral basic differences from the model of Saltzman and Kelso(1987), in addition to the differences with respect to the roleof physiological and psychological considerations in justify-ing the mathematics. The structure of Figure 12.7 is a kind ofcentral closed-loop system. This system, however, is inoper-ative as long as the Go signal is zero; it is energized by the Gosignal, which in addition can change across time so that thesystem is no longer linear. Bullock and Grossberg (1988)refer to a “factorization of pattern and energy.” Basically, theGo signal allows a separation of movement planningfrom movement initiation (cf. Gielen, van den Heuvel, &van Gisbergen, 1984), which implies that processes of motorpreparation can be temporally separated from execution ofthe movement, but also that movements can be initiatedbefore advance specification is finished.

Generative structures are not restricted to aimed move-ments. In fact, models of generative structures for periodicmovements as they occur in locomotion are historically older.Network models of central pattern generators had alreadybeen proposed early in the twentieth century (Brown, 1911),and more elaborate versions continue to be developed (e.g.,Grossberg, Pribe, & Cohen, 1997). In more abstract models,

of course, point attractors can be replaced by limit-cycle at-tractors which produce stable oscillations (e.g., Kay, Kelso,Saltzman, & Schöner, 1987).

The Advance Specification of Movement Characteristics

During motor preparation an anticipatory representation ofthe forthcoming movement is constructed. This representa-tion can be described as a motor-control structure, whichallows (relatively) autonomous control of the movement in-dependent of sensory feedback. In addition to being set up, thestructure must be specified, with the appropriate parameters.This is a time-consuming process. Thus, variations in neces-sary preparatory activities are reflected in reaction times. Inaddition, when the available time is varied, it is possible totrace the time course of the specification of movement char-acteristics. Thus far, almost all studies on the advance speci-fication of movement characteristics have employed aimedmovements or isometric contractions with different quantita-tive characteristics, yet qualitatively different movementshave hardly been used. Therefore, little can be said about set-ting up different motor-control structures, but more can besaid about the advance specification of parameters.

Figure 12.8 gives an example for the gradual specificationof movement direction, adapted from Georgopoulos, Lurito,Petrides, Schwartz, and Massey (1989). These data are froma monkey who had been trained to perform a movement toone of eight potential targets arranged on a circle. When thetarget was dimly illuminated, the monkey had to reach for itdirectly, but when the luminance of the target was high, themonkey had to perform a movement that was rotated by 90°counterclockwise relative to the target. What is shown inFigure 12.8 is the gradual rotation of the population vector in

T

V

P

�

�

�

�

�

� G

Figure 12.7 Variables of the VITE model of Bullock and Grossberg (1988)and their interrelations.

The Use of Sensory Information 327

such trials from the direction of the target (90°) to the direc-tion of the movement (180°). The population vector is com-puted from the activity of directionally tuned neurons of themotor cortex and generally points in the direction of move-ment. Basically it is a weighted mean of the preferred direc-tions of a sufficiently large sample of cortical units, with theweights being derived from the spike frequencies. The rota-tion of the population vector starts with a certain delay andproceeds with an almost constant slope until the target direc-tion is reached. In human subjects this kind of rotation pre-sumably gives rise to a systematic increase of reaction timewhen the angle between target and required direction ofmovement is increased (Georgopoulos & Massey, 1987).

The timed-response procedure allows one to trace thegradual specification of movement parameters from behav-ioral data. The method has been introduced for the study ofthe speed-accuracy trade-off in choice reaction time experi-ments (Schouten & Becker, 1967), and it has been adapted tothe study of the advance specification of characteristicsof isometric contractions and movements by Ghez andcoworkers (Ghez et al., 1997; Hening, Favilla, & Ghez,1988). Basically the method specifies a moment for the startof the movement; typically the movement has to be initiatedin synchrony with the last of four tones which are presentedin regular intervals. At a variable time before the last tone thetarget is presented, so the time available for motor specifica-tions can be varied. The method is only suited for rapidmovements or isometric contractions with short durations, sothat the movement characteristics are largely determined inadvance and little changed during execution.

Ghez and coworkers demonstrated the gradual specifica-tion of peak forces of isometric contractions as well asamplitudes and directions of movements with a time coursesimilar to that of the neuronal population vector (cf. Figure12.8). In addition, they showed that the gradual specificationsbreak down when the differences between the alternative tar-gets become too large (Ghez et al., 1997). When the differencebetween target directions is about 90° or larger, or the ratio oftarget amplitudes isabout12:1or larger, the intermediatevaluesbetween the two targets are no longer observed, and the choicebetween movement parameters becomes discrete. Thus, thereseem to be two qualitatively different modes of parameter spec-ification, namely gradual adjustments and discrete choices.

While the timed-response procedure provides a windowinto the gradual or discrete specification of movement char-acteristics, it has not been used as extensively as chronomet-ric procedures. The latter type of studies is largely basedon the movement precuing rationale of Rosenbaum (1980,1983). Consider a set of four responses that differ on two di-mensions like direction and amplitude. In a reaction time

task, before presentation of the response signal, there is thusuncertainty with respect to both direction and amplitude, andafter presentation of the response signal–during the reaction-time interval–both response characteristics have to be speci-fied. When one of the dimensions is precued, it can bespecified in advance of the response signal, and only onedimension remains to be specified after its presentation. Re-action time should be reduced by the time it takes to specifythe precued dimension. When both dimensions are precued,both can be specified in advance, and reaction time should bereduced even more. In principle, if the rationale were fullyvalid, the times needed to specify various movement charac-teristics or combinations thereof could be estimated.

There are some broad conclusions that can be drawn fromthe results obtained, but there are also a number of problemsthat sometimes cast doubt on the general validity of the ratio-nale (cf. Goodman & Kelso, 1980; Zelaznik, Shapiro, &Carter, 1982). Among the broad conclusions were that move-ment features are specified sequentially and in variable ratherthan fixed order (Rosenbaum, 1983). The first of these twobroad conclusions can be doubted because the time neededto specify two dimensions can be smaller than the sum ofthe times needed to specify each of these dimensions (e.g.,Lépine, Glencross, & Requin, 1989). In addition, timed-response studies show essentially parallel specifications ofamplitude and direction, perhaps accompanied by some slow-ing when two response characteristics are specified in parallel(Favilla & De Cecco, 1996; Favilla, Hening, & Ghez, 1989).

Exceptions to the second broad conclusion seem to be rare.Fixed order of specifications is indicated by a shortening of re-action time when a movement dimension A is precued, whichcan be observed only when movement dimension B is precuedas well, but not otherwise. This implies that the specificationof dimension B is a prerequisite for specifyingA. Such a resultwould be expected when dimension B embraces qualitativelydifferent movements, related to different motor-control struc-tures rather than to different parameters of a single controlstructure. Qualitative variations of movement characteristics,however, have rarely been studied, but some results of Roth(1988) indeed suggest that precuing the direction and theforce for throwing a ball does not result in systematic reactiontime benefits as long as the type of throw is not known.

THE USE OF SENSORY INFORMATION

The use of sensory information for the control of voluntarymovement was among the historically early questionsaddressed by experimental psychology. Woodworth (1899)asked his subjects to produce reciprocal movements between

328 Motor Control

Figure 12.9 Examples of handwriting with (upper example) and without(lower example) vision in (a) a deafferented patient (from Teasdale et al.,1993) and (b) a healthy girl.

two target lines in the pace of a metronome. With the partici-pants’ eyes closed, accuracy was only little affected byfrequency, but with the participants’ eyes open, accuracy in-creased relative to that found with closed eyes as soon as lessthan about two movements per second were produced. A nextmajor step was a study by Keele and Posner (1968) with dis-crete movements. Movement times were instructed, and themovements were performed with full vision or in the dark,with the room light being switched off at the start of themovements. Except for the shortest movement time of about190 ms, the percentage of movements that hit the targetwas larger with than without vision. Subsequent studiesshowed that the minimal duration at which accuracy gainsfrom the availability of vision becomes shorter—about100 ms—when conditions with and without vision areblocked rather than randomized (Elliott & Allard, 1985;Zelaznik, Hawkins, & Kisselburgh, 1983). This minimal du-ration reflects processing delays, but it also reflects the time ittakes until a change of the pattern of muscular activity has aneffect on the movement.

Woodworth (1899) distinguished between two phases of arapid aimed movement, an “initial adjustment” and a secondphase of “current control.” This distinction seems to implythat accuracy should profit mainly when vision becomesavailable toward the end of aimed movements. However,even early vision can increase accuracy (Paillard, 1982), andaccuracy increases when both initial and terminal periods ofvision increase in duration (Spijkers, 1993). Thus, the viewthat vision is important only in the late parts of an aimedmovement seems to be overly simplified.

From the basic findings it is clear that, in general, vision isnot really necessary for the production of movements, butthat it serves to improve accuracy. The same kind of general-ization holds for the second important type of sensory infor-mation for motor control, proprioception. (For tasks thatinvolve head movements, including stance and locomotion,the sensors of the inner ear also become important, althoughI shall neglect them here.) Regarding the role of propriocep-tion for motor control, classic observations date back toLashley (1917). Due to a spinal-cord lesion, the left kneejoint of his patient was largely anesthetic and without cuta-neous and tendon reflexes. In particular, the patient did notexperience passive movements of the joint, nor could he re-produce them; only fairly rapid movements were noted, butthe experienced direction of movement appeared random.However, when the patient was asked to move his foot by acertain distance specified in inches, the movements weresurprisingly accurate, as were the reproductions of activemovements; the latter reached the accuracy of a controlsubject. The basic finding that aimed movements are possible

without proprioception (and, of course, without vision also)has been confirmed both in monkeys (e.g., Polit & Bizzi,1979; Taub, Goldberg, & Taub, 1975) and—with local tran-sient anesthesia—in humans (e.g., Kelso & Holt, 1980),although, of course, without proprioception there tends to bea reduction of accuracy.

The very fact that movements are possible without visionand proprioception proves that motor control is not just aclosed-loop process but involves autonomous processes thatdo not depend on afferent information. The very fact thataccuracy is generally increased when sensory informationbecomes available proves that motor-control structures alsointegrate this type of information. Beyond these basic gener-alizations, however, the use of sensory information becomesa highly complicated research issue because sensory infor-mation can be of various types and serves different purposesin motor control.

As a first example of some complexities, consider a tasklike writing or drawing. Normally we have no problems writ-ing with our eyes closed, except that the positioning of theletters and words tends to become somewhat irregular in bothdimensions of the plane. This is illustrated in Figure 12.9b.Figure 12.9a shows the writing of a deafferented patient bothwith and without vision (Teasdale et al., 1993). The patienthad suffered a permanent loss of myelinated sensory fibersfollowing episodes of sensory neuropathy, which resulted in

The Use of Sensory Information 329

a total loss of sensitivity to touch, vibration, pressure, andkinesthesia as well as an absence of tendon reflexes, althoughthe motor nerve conduction velocities were normal. With vi-sion, the writing of “Il fait tiède” seems rather normal, butwithout vision the placement of words, letters, and parts ofletters is severely impaired, while individual letters remainlargely intact. Similarly, in drawing ellipses with eyes closed,single ellipses appeared rather normal, but successive el-lipses were displaced in space. Thus, absence of sensory in-formation affects different aspects of the skill differently, andimpairments are less severe when proprioception can serve asa substitute for absent vision.

Target Information

Vision and proprioception serve at least two different func-tions in motor control, which are not always clearly distin-guished. First, they provide information about the desiredmovement or target information, and, second, they provideinformation about the actual movement or feedback informa-tion. In the typical case, target information is provided by vi-sion only, and feedback information both by proprioceptionand by vision. Thus, vision provides both kinds of informa-tion, and the effects of absent vision can be attributed toeither of them. The obvious question of whether targetinformation or feedback information is more important formovement accuracy, as straightforward as it appears, cannotunequivocally be answered. In the literature, contrasting find-ings have been reported. For example, Carlton (1981) foundvision of the hand to be more important, whereas Elliott andMadalena (1987) found vision of the target to be crucial forhigh levels of accuracy. Perhaps the results depend on subtletask characteristics. However, for throwing-like tasks, visionof the target seems to be critical in general (e.g., Whiting &Cockerill, 1974), and dissociating the direction of gaze fromthe direction of the throw or shot seems to be a critical ele-ment of successful penalties.

Specification of Spatial Targets

Targets for voluntary movements are typically defined in ex-trinsic or extrapersonal space, whereas movements are pro-duced and proprioceptively sensed in personal space. Bothkinds of space must be related to each other; they must be cal-ibrated so that positions in extrinsic space can be assigned topositions in personal space and vice versa. When we movearound, the calibration must be updated because personalspace is shifted relative to extrinsic space. Even when we donot move around, the calibration tends to be labile. Thislability can be evidenced from the examples of handwriting

in Figure 12.9: With the writer’s eyes closed, calibration getslost with the passage of time, so positions of letters or parts ofthem exhibit drift or random variation. This effect is muchstronger when no proprioception is available.

An interesting example of failures that are at least partlycaused by miscalibrations of extrinsic and personal space areunintended accelerations (cf. Schmidt, 1989). These occur inautomatic-transmission cars when the transmission selectoris shifted to the drive or reverse position, typically when thedriver has just entered the car; when he or she is not familiarwith the car, this is an additional risk factor. In manual-transmission cars, incidents of unintended acceleration areessentially absent. According to all that is known, unintendedaccelerations are caused by a misplacement of the right footon the accelerator pedal rather than on the brake pedal with-out the driver’s being aware of this. Thus, when the car startsto move, he or she will press harder, which then has the un-expected effect of accelerating the car.

The position of the brake pedal is defined in the extrinsicspace of the car, whereas the foot placement is defined in thepersonal space of the driver. In particular upon entering a car,and more so when it is an unfamiliar car, there is the risk ofinitial miscalibration. Thus, when extrinsic and personalspace are not properly aligned, the correct placement of thefoot in personal space might reach the wrong pedal in extrin-sic space. Manual-transmission cars, in contrast, have a kindof built-in safeguard against such an initial miscalibration,because shifting gears requires that the clutch be operatedbeforehand. Thus, before the car is set into motion, the properrelation between foot placements and pedal positions isestablished.

Calibration, in principle, requires that objects, the loca-tions of which are defined in world coordinates, be simulta-neously located in personal space. Mostly it is vision thatserves this purpose. However, personal space embraces notonly vision: In addition to visual space, there are also a pro-prioceptive and a motor space, and these different spacesmust be properly aligned with each other. For example, inorder for us to reach to a visually located target, its locationmust be transformed into motor space, that is, into the appro-priate parameters of a motor control structure. In addition, itslocation must be transformed into proprioceptive space, sothat we can see and feel the limb in the same position. In alater section I shall discuss the plasticity of these relations;here I shall focus on the question of how a visually locatedspatial target is transformed into motor space.

An object can be localized visually both with respect to anobserver (egocentrically) and with respect to another object(allocentrically or exocentrically; cf. the chapter by Proffitt &Caudek in this volume). Geometrically the location of the

330 Motor Control

object can be described in terms of a vector. The lengthof the vector corresponds to the distance from the reference tothe object; for egocentric location the reference is a point be-tween the eyes (the cyclopean eye), and for allocentric locationit is another object in the visual field. The direction is usuallyspecified by angles both in a reference plane and orthogonal toit, but for the following its specification is of little importance.The available data suggest that both egocentric and allocentriclocalizations are used in the visual specification of targets.Which one dominates seems to depend on task characteristics.

Figure 12.10 shows a well-known optical illusion, theMüller-Lyer illusion. Although the length of the shaft is thesame in both figures, it appears longer in the figure with out-going fins than in the figure with ingoing fins. Elliott and Lee(1995) used one of the intersections as the start position andthe other intersection as the target position for aimed move-ments. Corresponding to the difference in perceived distancebetween the intersections in the two figures, movement am-plitudes were longer with outgoing fins than with ingoingfins (cf. Gentilucci, Chiefi, Daprati, Saetti, & Toni, 1996). Incontrast to this result, Mack, Heuer, Villardi, and Chambers(1985) found no effect or only a very small effect of theillusion on pointing responses.

Perhaps the critical difference to the study of Elliott andLee (1995) was that the participants in the study of Macket al. (1985) pointed not from one intersection to the other,but from a start position in their lap to one or the other of thetwo intersections. The difference between the two tasks sug-gests that the movements were based on allocentric (visualdistance) and egocentric (visual location) information, re-spectively. In fact, when psychophysical judgments of thelength of the shaft are replaced by judgments of the positionsof the intersections, the illusion also disappears (Gillam &Chambers, 1985). Thus, although physically a distance isthe difference between two positions on a line, this is notnecessarily true for perceived distances and positions. Thisdistinction between perception of location and perception ofdistance matches a distinction between different types of pa-rameters for motor control structures, namely target positionsversus distances (cf. Bock & Arnold, 1993; Nougier et al.,1996; Vindras & Viviani, 1998).

Specification of spatial targets in terms of distancesimplies a kind of relative reference system for a singlemovement: Wherever it starts, this position constitutes the

origin. A visually registered distance (and direction) is thenused to specify a movement in terms of distance (and direc-tion) from the start position. This way of specifying move-ment characteristics has a straightforward consequence:Spatial errors should propagate across a sequence of move-ments. In contrast, with a fixed reference system as impliedby the specification of target locations in terms of (egocen-tric) positions, spatial errors should not propagate. In studiesbased on this principle, Bock and Eckmiller (1986) and Bockand Arnold (1993) provided evidence for relative referencesystems, that is, for amplitude specifications. The movementsthey studied were pointing movements with the invisiblehand to a series of visual targets. However, Bock and Arnoldalso noted that error propagation was less than perfect. Heuerand Sangals (1998) used different analytical procedures, butthese were based on the same principle of error propagationor the lack thereof. As would be expected, when only ampli-tudes and directions were indicated to the subjects, only a rel-ative reference system was used. However, when sequencesof target positions were shown, there was some influence of afixed reference system, although the movements were per-formed on a digitizer and thus displaced from the targetpresentation in a manner similar to the way a computermouse is used.

Gordon, Ghilardi, and Ghez (1994) provided evidence fora reference system with the origin in the start position basedon a different rationale, again with a task in which targetswere presented on a monitor and movements were performedon a digitizer. Targets were located on circles around the startposition. The distribution of end-positions of movements to asingle target typically has an elliptical shape. Under the as-sumption that the target position is specified in terms of di-rection and distance from the origin of the reference system,the axes of the elliptical error distributions, determined byprincipal component analysis, should be oriented in a partic-ular way: The axes (one from each endpoint distribution)should cross in the origin. It turned out that the long axes ofthe error ellipses all pointed to the start position, as shownin Figure 12.11. Corresponding findings were reported byVindras and Viviani (1998), who kept the target position con-stant but varied the start position.

Amplitude specifications allow accurate movements evenwhen visual space and proprioceptive-motor space are notprecisely aligned. Specifically, they do not require absolutecalibration, but only relative calibration: It must be possibleto map distances correctly from one space to another, but notpositions. Of course, without absolute calibration, move-ments may drift away from that region of space where thetargets are, as is typical with the use of a computer mouse.Without proprioception it seems that absolute calibration isessentially missing. In the case of the deafferented patient

Figure 12.10 The Müller-Lyer illusion.

The Use of Sensory Information 331

Figure 12.11 Elliptical end-position distributions of movements from astart position to concentrically arranged targets; circles mark the target areas(after Gordon et al., 1994).

mentioned above, Nougier et al. (1996) found basically cor-rect amplitude specifications in periodic movements betweentwo targets, although there were gross errors in the actualend-positions relative to the targets.

Contrasting with the evidence for amplitude specificationsor relative reference systems in tasks of the type “reachingfrom one object to another,” in tasks of the type “reaching outfor an object” there is evidence for a reference system that isfixed, with the origin being at the shoulder or at a location in-termediate between head and shoulder (Flanders, HelmsTillery, & Soechting, 1992). The analyses that led to this con-clusion were again based on the assumption that errors of am-plitude and direction should be essentially independent.However, when the start position of the hand is varied, an in-fluence can again be seen, but not as dominant an influence asin the task of Gordon et al. (1994). Thus, McIntyre, Stratta,and Lacquaniti (1998) concluded that there is a mixtureof different reference systems; in addition, errors of visuallocalization are added to errors of pointing.

Taken together, the evidence suggests that target infor-mation in general is specified both in terms of (egocentric)positions and in terms of (allocentric) distances and direc-tions. Localization in terms of egocentric positions requiresthat, to perform a movement, the visual reference systembe transformed to a proprioceptive-motor reference system,the first having its origin at the cyclopean eye, the latter hav-ing its origin at the shoulder, at least for certain types of armmovements. Localization in terms of allocentric distancesand directions requires that the visual reference system be

aligned with the proprioceptive-motor reference system in away that the origin is in the current position of the end-effector. The relative importance of the two reference sys-tems depends on task characteristics. In addition, there isalso evidence that it can be modulated intentionally(Abrams & Landgraf, 1990).

Although spatial targets are mostly specified visually, theycan also be specified proprioceptively, and again there is evi-dence for target specifications in terms of both position andamplitude, with the relative importance of these being af-fected both by task characteristics and intentions. In these ex-periments, participants produce a movement to a mechanicalstop and thereafter reproduce this movement. When the startposition is different for the second movement, participantscan be instructed to reproduce either the amplitude of the firstmovement or its end-position. The general finding is a biastoward the target amplitude when the task is to reproduce theend-position, and a bias toward the end-position when thetask is to reproduce the amplitude (Laabs, 1974). Althoughtypically the reproduction of the end-position is more accu-rate than the reproduction of the amplitude, this is more so forlonger movements, less so for shorter ones, and it may evenbe reversed for very short ones (Gundry, 1975; Stelmach,Kelso, & Wallace, 1975).

Specification of Temporal Targets

In tasks like catching, precisely timed movements are re-quired: The hand must be in the proper place at the propertime and be closed with the proper timing to hold the ball. Invery simple experimental tasks, finger taps have to be syn-chronized with pacing tones. Although the specification oftemporal targets is fairly trivial in such tasks, the findings re-veal to which aspects of the movements temporal goals arerelated. A characteristic finding is negative asynchrony, a sys-tematic lead of the taps in the range of 20–50 ms, which, forexample, is longer for tapping with the foot than for tappingwith the finger (e.g., Aschersleben & Prinz, 1995).

The negative asynchrony is taken to indicate that the tem-poral target is not related to the physical movement itself, butrather to its sensory consequences, proprioceptive and tactileones in particular, but also additional auditory ones if they arepresent. For example, because of the longer nerve-conductiontimes, sensory consequences of foot movements should becentrally available only later than sensory consequences ofhand movements; thus negative asynchrony is larger in theformer case than in the latter. When auditory feedback isadded to the taps, negative asynchrony can be manipulated byvarying the delay of the auditory feedback relative to the taps(Aschersleben & Prinz, 1997): Negative asynchrony declineswhen feedback tones are added without delay and increases

332 Motor Control

Figure 12.12 (a) Spatial layout of a simple interceptive task. (b) Variablesin the analysis of the task; time zero is defined by the target’s reaching theintersection.

as the delay becomes longer. With impaired tactile feedback,sensory consequences should also be delayed centrally, andnegative asynchrony is increased (Aschersleben, Gehrke, &Prinz, 2001).

Synchronization of movements with discrete tones isnecessarily anticipatory, provided that the interval betweensuccessive tones is sufficiently short (Engström, Kelso, &Holroyd, 1996). This is different in interceptive tasks. For ex-ample, when an object is approaching and one has to performa frontoparallel movement that reaches the intersection of theobject path and the movement path at the same time as theobject does (cf. Figure 12.12a), it is possible in principle tocontinuously adjust the distance of the hand from the inter-section to the distance of the object. In fact, this may actuallyhappen if both the target object and the hand move slowly. Atleast, it is true that slower movements are adjusted more ex-tensively to the approaching target after their start than rapidmovements.

Let the start time be the time interval between the start ofthe interceptive movement and the time the target objectreaches the intersection, and the temporal error be the timebetween the hand’s and the target object’s reaching the inter-section (Figure 12.12b). Then, when the movement is startedand runs off without further adjustments of its timing, thestart time should be highly correlated with the temporal error.This strategy, in which the start time is selected according to

the expected duration of a pre-selected movement pattern, issometimes called operational timing (Tyldesley & Whiting,1975). However, with temporal adjustments the correlationbetween start time and error should be reduced (Schmidt,1972). This happens when the instructed movement durationis increased (Schmidt & Russell, 1972). Thus it seems that onthe one hand the interceptive movement can be triggered by aparticular state of the approaching object and then run offwithout further adjustments, and that on the other hand thetime course of the interceptive movement can be guided bythe approaching object, with mixtures of these two modesbeing possible.

In the simple task considered thus far the position of theintersection of object path and hand path is given. This is dif-ferent for more natural tasks. Consider hitting a target thatmoves on a straight path in a frontoparallel plane like a spideron the wall. In principle, spiders can be hit in arbitrary places,but nevertheless the direction of the hitting movement has tobe adjusted to an anticipated position of the moving target. Arobust strategy is to adjust the lateral position of the hand tocontinuously updated estimates of the target position at thetime the hand will reach the target plane; this requires an esti-mate of the time that remains until the hand reaches the planeand an estimate of the target’s velocity, which, however, neednot really be correct (Smeets & Brenner, 1995).

The situation is somewhat different when balls have to beintercepted in a lateral position, either for catching them orfor hitting them. According to Peper, Bootsma, Mestre, andBakker (1994), the hand will be in the correct position in theplane of interception at the right time when its lateral veloc-ity is continuously adjusted to the current difference betweenthe lateral position of the hand and the approaching target, di-vided by the time that remains until the target reaches theplane of intersection. Proper lateral adjustments, which implytemporal adjustments as well, are evident even in high-speedskills like table tennis, although the relevant information isless clear (Bootsma & van Wieringen, 1990).

What is the basis for anticipations of temporal targets? Forexample, when we view an approaching ball, what allows usto predict when it will be in some position where we canintercept it (cf. the chapter by Proffitt & Caudek in thisvolume)? The time it takes until a moving object reaches acertain position is given by the distance of the object dividedby its velocity. This ratio has time as unit, and it specifies timeto contact with the position, provided the object moves on astraight path with constant velocity. As noted by Lee (1976),the information required to determine time to contact with anapproaching object, or with an object the observer is ap-proaching, is available even without determining distance andvelocity, namely by the ratio of the size of the retinal image ofthe object and its rate of change. This variable, called �, has

The Use of Sensory Information 333

become quite popular. There can be little doubt that it con-tributes both to temporal judgments (e.g., Schiff & Detwiler,1979) and precisely timed actions (e.g., Savelsbergh,Whiting, & Bootsma, 1991). However, it is not the only rele-vant information; other kinds of information, for examplebinocular distance information, are used as well (Bennett, vander Kamp, Savelsbergh, & Davids, 1999; Heuer, 1993a). In arecent overview, Tresilian (1999) notes that the relation be-tween rapid interceptive actions and the kind of informationused is rather flexible and in no way invariant. There is adegree of task dependence that at present does not allow firmgeneralizations about how rapid interceptive actions areadjusted to their temporal targets.

Feedback Information

Although movements can be performed in the absence of af-ferent information from the moving limb with an astonishingdegree of accuracy, the use of feedback information isindicated by the effects of perturbations of feedback on per-formance. For example, proprioceptive information can bedistorted by way of tendon vibration with a vibrator placed inthe proper position on the skin. The effect is a tonic excitationof muscle spindles, which under normal conditions corre-sponds to a longer muscle and correspondingly different jointangle. If, for example, the biceps tendon is vibrated, theelbow angle is registered as being too large. When the elbowangle has to be matched to the elbow angle of the other arm,the matched angle is too small, corresponding to the dis-torted proprioceptive feedback on joint angles (Goodwin,McCloskey, & Matthews, 1972).

Regarding the effects of distorted visual feedback, a par-ticularly striking example has been reported by Nielsen(1963). The participant’s task was to move one hand along avertical line, but the visible gloved hand was that of the ex-perimenter and followed a curved path rather than a straightone. Subjects attempted to correct the error so that they devi-ated from the target line in the opposite direction. In spite ofthe strong discrepancies between intended and felt movementon the one hand and visual feedback on the other hand it tookseveral trials before participants came to realize that thevisible gloved hand could not be their own.

In simple movements, feedback information is function-ally of little importance because autonomous processes ofmotor control can operate on the basis of a sufficiently accu-rate internal model of the motor transformation, so that onlylittle error remains for closed-loop control to operate on (ex-cept, of course, when feedback information is distorted).However, in tasks in which a sufficiently accurate internalmodel is not available, the availability of visual feedbackgains critical importance. This is the case when we operate

sufficiently complex machines or tools which effectively addto the normal motor transformation. Experimentally trackingtasks are suited to exploring the role of visual feedback(Poulton, 1957).

For example, when the movement of the hand is propor-tional to the motion of the cursor on a screen, trackingperformance is rather robust against short periods of elimi-nated visual feedback. However, with velocity control–withwhich the position of the hand is proportional to the veloc-ity of the cursor on the screen–even short periods of elimi-nated feedback can bring performance down to an almostchance level (e.g., Heuer, 1983, p. 54). Thus, visual feed-back gains in importance the less accurate the internalmodel of the transformation by a machine is. Internal mod-els of sufficiently complex transformations seem not to bedeveloped, so that practice does not reduce the critical im-portance of visual feedback (Davidson, Jones, Sirisena, &Andreae, 2000).

Feedback information does not only serve to guide an on-going movement; it is also required to learn and to maintainan internal model of a transformation (cf. Jordan, 1996), pro-vided it is not too complex. For example, Sangals (1997) hadhis subjects practice a nonlinear relation between the ampli-tude of the movement of a computer mouse and the amplitudeof the cursor movement. When visual feedback during eachmovement of a sequence was eliminated and only terminalfeedback at the end of each movement was provided, the re-lation between (visual) target amplitudes and movement am-plitudes remained nonlinear. However, when visual feedbackwas completely eliminated for a sequence of several move-ments, the relation between (visual) target amplitudes andmovement amplitudes became linear, which is likely to be akind of default relation (cf. Koh & Meyer, 1991).

Feedback information can be processed at various levelsof control; that is, it can be integrated with autonomous con-trol processes in different ways. Consider the simple task ofsine tracking: Subjects control the motion of a cursor on thescreen, with the target following a sinusoidal time course. Inprinciple, subjects could function like a position servo, mini-mizing the deviation between the position of the cursor andthe position of the target. In fact, with a low frequency of tar-get motion this may actually be the case. However, withhigher frequencies, which approach the range where perfor-mance breaks down, human subjects produce a sinusoidalmovement and seem to adjust its frequency and phase(Noble, Fitts, & Warren, 1955). Similar indications for theprocessing of parametric feedback rather than positionalfeedback have been reported by Pew (1966), but in generalthe processing of parametric rather than positional feedbackhas received very little attention. In everyday tasks likedriving a car it may be of critical importance; perhaps it is not

334 Motor Control

the position of the car on the road that is controlled but,rather, parameters like the curvature of the path.

In tasks like tracking there is not only visual feedback, butalso proprioceptive feedback, and both types of feedback are re-lated to different objects: proprioceptive feedback to one’s ownmovements, but visual feedback to the motions of a controlledobject. The relation between both types of feedback depends onthe transformation implemented by the controlled machine.Only when the transformation is simple or well-learned, or both,can proprioceptive feedback replace visual feedback, but ingeneral such intermodal matching is associated with some lossof accuracy as compared with intramodal matching of target andfeedback information (e.g., Legge, 1965). On the other hand,when visual and proprioceptive feedback are different butnevertheless refer to the same object, as in the classical task ofmirror drawing, the absence of proprioceptive feedback can ac-tually enhance performance (Lajoie et al., 1992).

Even when visual and proprioceptive feedback refer tothe same object, they are not necessarily redundant. For ex-ample, when we use a knife, both vision and proprioceptionprovide information about its current position with respect tothe object to be cut. Of course, visual information is moreaccurate in this respect, and as far as the spatiotemporalcharacteristics of the movement are concerned, propriocep-tive information is not really needed. However, it providesinformation that is not available visually, in particular aboutthe resistance of the cut object. Thus, although vision is crit-ical for registering spatiotemporal characteristics, proprio-ception is critical for registering force characteristics. Thelack of this latter kind of information is a problem in remotecontrol and other tasks that followed from recent technolog-ical developments (cf. the chapter by Klatzky & Ledermanin this volume).

An example is minimally invasive surgery (cf. Tendick,Jennings, Tharp, & Stark, 1993). Such operations are per-formed by means of an endoscope and instruments that arepivoted roughly at their place of insertion into the tissue.Although the facts that movements of the hand result inmovements of the tip of the instrument in the opposite direc-tion and that the gain of lateral movements depends on trans-lational movements seem not to pose severe practicalproblems, the lack of appropriate force feedback seems to bemore critical. In particular, there is only poor proprioceptiveinformation about reactive forces at the tip of the instrument,so there is the risk of damage to the tissue operated on.

Sensory Information for Motor Control and Perception

Much of the sensory information that is involved in the con-trol of movements apparently has no access to consciousness.

Folklore knows that one just has to do it without attending toomuch to how it is done. In fact, Wulff, Höß, and Prinz (1998)found better learning of gross motor skills when the attentionof the learners was focused on the effects of the movementsrather than on the movements themselves, for example on astabilometer platform rather than on the feet (for review, seeWulf & Prinz, 2001). It is not only that we do not perceive ourmovements in all details—for example, in skills like the longjump we do not normally perceive the details of the move-ments of our extremities (Voigt, 1933)—but, in addition, ourmovements can be more precise than would be expectedfrom the limits of our perceptual skills. This was not only oneof the major claims of a motor branch of the so-calledGanzheitspsychologie (Klemm, 1938), but it has also beenemphasized in more recent times. For example, McLeod,McLaughlin, and Nimmo-Smith (1985) ascribed the verysmall temporal variability in batting of only a few millisec-onds to the functioning of a dedicated special-purpose mech-anism. In any case, hitting a falling ball at a certain positionof its path is more precise than pressing a key when the ballreaches the same position (Bootsma, 1989).

Clinical cases illustrate that humans can reach to visualtargets that they do not perceive, provided that the blind areasof the visual field (scotoma) are caused by certain lesions(e.g., Campion, Latto, & Smith, 1983; Perenin & Jeannerod,1978). This phenomenon has become known as blindsight. Inaddition, clinical data give evidence of double dissociations.For example, some patients can identify and describe objects,but they cannot use the information about size, form, andorientation of the objects to grasp them; other patients, incontrast, cannot perceive these features of objects, but never-theless can grasp them (Goodale & Milner, 1992).

The dissociability of visual information for perception andfor motor control supports a theoretical distinction that hasreceived much attention during the last 20 years. Basing theiridea mainly on lesion studies, Ungerleider and Mishkin(1982) proposed the distinction between two cortical visualsystems, one involving the inferotemporal cortex and theother the posterior parietal cortex (ventral stream and dorsalstream, respectively). In functional terms, these two systemshave been characterized as the what system and the wheresystem, the former serving object identification and the latterspace perception. Alternatively, they are characterized func-tionally as the what system and the how system, the formerbeing in the service of perception and the latter in the serviceof motor control (e.g., Goodale & Humphrey, 1998; Goodale& Milner, 1992). This latter functional characterization doeslargely coincide with a distinction between a cognitive and asensorimotor system (Bridgeman, Kirch, & Sperling, 1981;Bridgeman, Lewis, Heit, & Nagle, 1979).

Motor Coordination 335

The evidence for the functional separation of a cognitiveand a sensorimotor system is based on differences betweenpsychophysical judgments and motor responses to identicalstimuli. For example, Bridgeman, Peery, and Anand (1997)exploited the long-known effect of asymmetric stimuli in thevisual field on the perceived direction of a target. They pre-sented targets within a frame which was centered or shifted tothe left or to the right. The target could appear in five differ-ent positions, and participants had to give their judgments bypressing one of five keys immediately after the stimulus haddisappeared. For these perceptual judgments there was aclear effect of the position of the frame: When the frame wasshifted to the left, judgments were shifted to the right, andvice versa. In contrast, when participants had to rotate apointer so that it pointed to the target just presented, abouthalf the participants exhibited no effect of the position of theframe. This was so although the response mode varied ran-domly and was cued only after the target had disappeared.

When delays of a few seconds between the disappearanceof the target and the response were introduced, all partici-pants showed effects of the frame position on pointing.Bridgeman et al. (1997) took their findings to indicate that thesensorimotor representation of the target is short-lived andoverridden by the cognitive representation when the delaybetween disappearance of the target and response becomessufficiently long. In some subjects the sensorimotor represen-tation might even be so short-lived that it hardly survives thetarget presentation.

Although the what versus how distinction currently has adominant influence, it is most likely a simplification. Pro-cessing of visual information is widely distributed acrossthe brain, and so is motor control. Thus, it is easy to conceiveof a set of systems that for different kinds of responses makeuse of different combinations of the various neural represen-tations of the visual world. From such a perspective, therewould be a multiplicity of perception-action systems, forwhich there is indeed evidence in other primates than humans(cf. Goodale & Humphrey, 1998).

MOTOR COORDINATION

In a general sense, coordination is a characteristic of almostany skilled movement, in that skilled performance requiresfairly precise relations between various kinematic, kinetic,and physiological variables. For example, in cranking (andrelated tasks like pedaling), force pulses need to be preciselytimed to occur during a certain phase of the rotation of thecrank or pedal (cf. Glencross, 1970); in rapid finger tapping,muscle activity of flexors and extensors must be timed to

occur at certain phases of the movement cycle (Heuer, 1998);in reaching for an object, the opening of the fingers must berelated to the movement of the hand toward the object (e.g.,Jeannerod, 1984); and so on. With this broad meaning, theterm coordination becomes almost equivalent to motor con-trol. However, for this section I use a narrower meaning inthat I focus on the coordinated movements of different effec-tors, mainly the two arms (interlimb coordination).

Task Constraints and Structural Constraints

Coordinated movements of the two hands are largely deter-mined by the task constraints. For example, the coordinationpattern for sweeping with a broom is different from that forbathing a baby. This certainly is not a fact that deserves elab-oration. However, there are more subtle consequences of taskconstraints. Perhaps the most important of these is compen-satory covariation.

As an example, consider the lip aperture in speaking. Aparticular lip aperture can be achieved by various combina-tions of the positions of the upper lip, the lower lip, and thejaw. These positions exhibit compensatory covariation suchthat, for example, a high position of the upper lip will be ac-companied by higher positions of the lower lip and/or thejaw, and a low position of the upper lip by lower positions ofthe lower lip and/or the jaw (Abbs, Gracco, & Cole, 1984).Compensatory covariation can be observed not only when lippositions vary spontaneously, but also when they are per-turbed by means of some mechanical device (Kelso, Tuller,Vatikiotis-Bateson, & Fowler, 1984). A task similar to reach-ing a certain lip aperture is that of grasping an object with aprecision grip, wherein there is compensatory covariation ofthe positions of thumb and index finger (Darling, Cole, &Abbs, 1988).

Compensatory covariation can be seen as a way to reducethe degrees of freedom in motor control. In addition, com-pensatory covariation contributes to solving the problem ofachieving a stable movement outcome in spite of variablecomponents. For example, with the appropriate covariationof lip and jaw positions, lip aperture will hardly vary. In fact,the principle of compensatory covariation seems to be a gen-eral principle of stabilizing movement outcomes (Müller,2001) which is not restricted to tasks in which different limbsare involved.

A highly illustrative task for compensatory covariations isthrowing a ball a certain distance. For physical reasons, whenthe initial flight angle varies, the initial velocity of the ballhas to covary to reach a certain target distance. In particular,with an initial angle of 45° the initial velocity has to be small-est, and it has to be increased as the initial flight angle

336 Motor Control

Figure 12.13 Equal-outcome curve for a particular dart-throwing task.When the target is an area rather than a point, deviations from the curve arepermitted, so that it becomes an area in which initial velocity and flight angleof successful throws must be located (after Müller, 2001).

deviates from 45°, provided that the height at which the ballis released is also the height at which the target is located.Conforming to these task constraints, Stimpel (1933) ob-served positive correlations across series of throws to a cer-tain target position between initial velocity and the absolutedeviation of the initial flight angle from 45°. It is not fullyclear how the particular covariation is established, but thereis the possibility that subjects produce equifinal trajectoriesof the hand such that across time initial flight angle and ve-locity covary in the proper way; thereby the proper relation isestablished independent of the precise time at which the ballis released (Müller & Loosch, 1999).

The plot of the relation between initial velocities and ini-tial flight angles required for a certain outcome of the throwscan be thought of as an equal-outcome curve (Heuer, 1989).Figure 12.13 gives an example for a particular dart-throwingtask, which also shows that with a target of a certain width,deviations from the bull’s-eye-outcome curve have differentconsequences for accuracy depending on where on the curvesubjects operate. In principle, equal-outcome curves can bedetermined for all sorts of tasks and for various sets of com-ponent variables. They specify how components of a skillmust be related to each other in the service of satisfying thetask constraints. In fact, this kind of analysis can becomeconsiderably more complex than what can be represented interms of equal-outcome curves (or perhaps areas). An exam-ple is the analysis of juggling by Beek (1989).

The very fact that components of a motor pattern covary inthe service of achieving particular outcomes has been taken asevidence for the existence of movement Gestalts (Bewegungs-gestalten) by Stimpel (1933) and his advisor Klemm (1938),

a notion that is analogous to perceptual Gestalts (see the chap-ter by Palmer in this volume). The core of the notion of aBewegungsgestalt is the idea that the whole dominates itscomponents and is more precise than expected from the com-ponents’ variabilities. Although these notions appear fairlyoutdated now, it cannot be overlooked that they anticipatesynergetic concepts (Haken, 1982) that currently play an im-portant role in the study of motor coordination (cf. Kelso,1994; Schöner, 1994). One of the core concepts is that of anorder parameter (or collective variable) that enslaves the com-ponent variables, so that higher level variables are not simplythe result of lower level variables, but dominate the lowerlevel variables instead. This general idea is captured by mod-els like the task-dynamic model of Saltzman and Kelso (1987)and the knowledge model of Rosenbaum, Loukopoulos,Meulenbroek, Vaughan, and Engelbrecht (1995).

Coordination in the service of satisfying task constraints isflexible: That is, patterns of covariation between certain ef-fectors that can be observed when one task is performed maybe absent when a different task is performed (e.g., Kelsoet al., 1984). Nevertheless, for biologically important taskslike standing, locomotion, eating, and so on, there may bemore rigid coordination patterns that not only support thesetasks, but may also impede performance of sufficiently dif-ferent tasks. Although it is not certain that such more rigidcoordination patterns for biologically important tasks are in-deed the origin of structural constraints on coordination, it iscertain that structural constraints do exist. Basically, theylimit the range of task-specific coordination patterns; whilethey support the production of certain patterns, they tend toimpede the production of deviating patterns.

Structural constraints support symmetrical movements ofthe two arms. Thus, mirror writing with the left hand be-comes a fairly simple task when it is performed concurrentlywith normal writing of the right hand (Jung & Fach, 1984).The other side of the coin is the difficulty we encounterwhen we attempt to produce different spatiotemporal patternsconcurrently with the two hands. Although it is certainly nottrue that both hands are constrained to act as a unit in thesense of having a common timing (Kelso, Southard, & Good-man, 1979; Schmidt et al., 1979), bimanual movements tendtoward identical durations, and only with strictly requireddifferent target durations can this tendency be overcome(Spijkers, Tachmatzidis, Debus, Fischer, & Kausche, 1994).Other deviations from strict symmetry are easier to achieve,but nevertheless there is a widespread tendency for dif-ferent movements with the two hands not to be as differentas they should be; the systematic errors here point to thesymmetric patterns that are the ones supported by structuralconstraints.

Motor Coordination 337

Figure 12.14 Potentials V(�) as specified by Haken et al. (1985) for dif-ferent ratios of b�a of the parameters.

The nature of structural constraints on coordination andtheir combination with task constraints or task-related inten-tions is nicely captured by perhaps the most influential modelof motor coordination and its developments (Haken, Kelso, &Bunz, 1985). This model applies to tasks that require concur-rent oscillations of at least two effectors, for example, the twohands. When these oscillatory movements are produced sym-metrically, there is essentially no difficulty in speeding themup as far as this is possible. However, when they are producedasymmetrically, the phase relation between the oscillations isless stable, and occasionally symmetric movement cycles in-trude (Cohen, 1971). When the asymmetric movements arespeeded up, stability is reduced even more, provided that sub-jects are instructed to maintain the asymmetric phase relation(Lee, Blandin, & Proteau, 1996). However, with a “let it go”instruction, subjects tend to switch to symmetric movementsat a certain critical frequency (Kelso, 1984).

Haken et al. (1985) modeled these phenomena in terms ofwhat came later to be called an intrinsic coordination dynam-ics. The model was formulated at two levels, the level of actualmovements and the level of an order parameter that capturesthe relation between the periodic movements. Basically, at thekinematic level two nonlinear oscillators were posited, one foreach effector, with a nonlinear coupling in addition. Relativephase � was chosen as the order parameter (or collective vari-able); this is the phase difference between the two oscillatorymovements. For this variable the dynamics were specifiedbased mainly on formal considerations: �

�= –a sin � –

2b sin 2�. Better known is the formulation in terms of a po-tential function V with �

�= dV�d�, V = –a cos � – b

cos 2�. This potential, which is illustrated in Figure 12.14, hasstable equilibria at � = n�, n = 0, �1, �2, . . . , provided theparameters a and b are within certain ranges. Stable equilibriaare characterized by �

�’s being positive for smaller values of �

and negative for larger values, so that relative phase will driftback to the equilibrium angle whenever it deviates as a conse-quence of some perturbation; in the potential function, stableequilibria are characterized by minima.

The ratio of the parameters a and b is hypothesized to de-pend on movement frequency, b becoming relatively smalleras frequency increases. When it becomes sufficiently small,the stable equilibria at � = m�, m = �1, �3, �5, . . . disap-pear (cf. Figure 12.14). This corresponds to the observationthat, as the frequency increases, only symmetric oscillations(in-phase oscillations in formal terms) are maintained whileasymmetric oscillations (anti-phase oscillations) tend toswitch to symmetric ones.

This account of the switch is based pretty much on formalconsiderations, and other models are available with strongerreference to physiological or psychological considerations or

both (Grossberg et al., 1997; Heuer, 1993b). In addition, theprediction that the switch should be associated with reducedmovement amplitudes (Haken et al., 1985) is not necessarilycorrect (Peper & Beek, 1998). Nevertheless, the model cap-tures nicely the soft nature of structural constraints, and anextension of it illustrates how structural constraints bias per-formance when they deviate from task constraints or task-related intentions.

Yamanishi, Kawato, and Suzuki (1980) asked their sub-jects to produce bimanual sequences of finger taps at variousphase relations. The stability of phasing was highest withsynchronous taps (relative phase of 0°), second highest withalternating taps (relative phase of 180°), and lower at all otherrelative phases. In addition the mean relative phases were bi-ased toward the stable relative phases at 0° and 180°. Schönerand Kelso (1988) modeled these effects by way of adding aterm to the intrinsic dynamics that reflects the “intention” toreach a target relative phase �, so that the potential becomesV = – a cos � – b cos 2� – c cos((� – �)�2). When theintended relative phase � differs from � = 0° and � = 180°,the minima of this potential are broader, corresponding toan increased variability, and shifted away from the intendedrelative phase, corresponding to the observed systematicbiases.

Basic Structural Constraints on Coordination

Structural constraints on coordination are indicated by sys-tematic errors. They have been studied mainly by means of

338 Motor Control

two different types of task, the one involving sequences ofmovements with different effectors and the other discretemovements, mostly of short duration. By and large thereseem to be no striking discrepancies between the conclusionsbased on the two types of experimental paradigm, althoughthe precise relation between them is not fully clear. For ex-ample, one cannot exclude that certain constraints mayevolve gradually so that they are effective for sequences ofmovements, but not for brief discrete ones.

A highly consistent finding is that periodic oscillations ofthe upper limbs are more stable when they are performedsymmetrically than when they are performed asymmetrically.The same kind of observation has also been made for bi-manual circle drawing (e.g., Carson, Thomas, Summers,Walkers, & Semjen, 1997). However, the symmetry con-straint, which favors the concurrent activation of homologousmuscle groups, is not universal; in addition, there is a bias to-ward identical movement directions (e.g., Serrien, Bogaerts,Suy, & Swinnen, 1999). This is particularly obvious for peri-odic movements with nonsymmetric effectors. For example,Baldissera, Cavallari, and Civaschi (1982) and Baldissera,Cavallari, Marini, and Tassone (1991) found that concurrentup-and-down movements of foot and hand in identical direc-tions are more precisely coordinated than concurrent move-ments in opposite directions. Thus, although essentially thereis always a preferred phase relation for periodic movementsof different effectors, which phase relation this is depends onthe particular effectors chosen and their plane of motion.

A second highly consistent finding is related to the timingof bimanual response sequences. Such sequences are simplewhen they have the same frequency, and they are also fairlyaccurately produced when the frequencies are harmonicallyrelated, that is, by integer ratios. However, for polyrhythmsperformance deteriorates (e.g., Klapp, 1979). For this it is notessential that the polyrhythms are produced by the two hands,but poor or even chance performance can also be observedin vocal-manual tasks (Klapp, 1981). There seems to be ageneral rule that the variability of the temporal errors of indi-vidual responses increases as the product mn for m : nrhythms increases (Deutsch, 1983); for example, variabilityis higher with a 2 : 5 (mn = 20) than with a 2 : 3 (mn = 6)rhythm. When the overall rate of polyrhythms is increased,not only does performance become poorer, but in additioncomplex rhythms may switch to less complex ones, like 2 : 5to 1 : 2 (e.g., Peper, 1995).

The observations on polyrhythms have been taken to sug-gest the existence of a unitary timing-control mechanism formovements of the two hands (Deutsch, 1983). In fact, formalanalyses of polyrhythm production in terms of timer models

(cf. Vorberg & Wing, 1996) generally reveal integratedcontrol, in which the timing of a response with the onehand can be relative to a preceding response with the otherhand (Jagacinski, Marshburn, Klapp, & Jones, 1988; Klappet al., 1985; Summers, Rosenbaum, Burns, & Ford, 1993).The difficulty in the production of polyrhythms is then basi-cally related to the complexity of the integrated timingcontrol structure. Only recently evidence has been reportedaccording to which professional pianists can exhibit paralleltiming control for the two hands when they producepolyrhythms at rapid rates (Krampe, Kliegl, Mayr, Engbert,& Vorberg, 2000). Except for such a select population, how-ever, temporal coupling appears to be so tight that tasks thatapparently require decoupling are performed in a way thatmaintains a unitary timing control.

Relatively little research effort has been invested in thestudy of sequences of bimanual movements with differentamplitudes. Franz, Zelaznik, and McCabe (1991) studied theconcurrent production of periodic lines and circles with thetwo hands. Drawing circles with one hand requires periodicoscillations with the same amplitudes along both axes of theplane, while drawing lines with the other hand requires a pe-riodic oscillation along only one axis; however, for the otheraxis, one can think of a periodic oscillation with zero ampli-tude. Franz et al. found that both the lines and the circles be-came elliptical (Figure 12.15). Thus, the different amplitudesof the oscillations became more similar in the bimanual task:The larger amplitude oscillations in drawing circles were re-duced in amplitude, and the zero-amplitude oscillations indrawing lines were enhanced in amplitude. More straightfor-wardly, such an amplitude assimilation for periodic move-ments was shown by Spijkers and Heuer (1995) and by Franz(1997), both for lines (that is, one-dimensional oscillations)and for circles (that is, two-dimensional oscillations). In ad-dition, Spijkers and Heuer (1995) found that the amplitudeassimilation became stronger as the frequency of oscillationswas increased.

Kelso, Tuller, and Harris (1983) had their subjects oscil-late a finger and concurrently repeat the syllable stock. Whenevery second syllable had to be stressed, there was an invol-untary increase of the amplitude of the accompanying fingermovement; similarly, voluntarily increased finger amplitudeswere accompanied by involuntary stresses of the syllable.These findings, which have been confirmed by Chang andHammond (1987), suggest that, in addition to amplitudeassimilation, changes of amplitude might overflow to othereffectors.

For a more systematic exploration of the contralateral ef-fects of large-to-small or small-to-large amplitude changes,

Motor Coordination 339

single-hand dual-handsame

Lines(a)

dual-handdifferent

2.5 cm

2.5 cm

single-hand

Circles(b)

dual-handsame

dual-handdifferent

2.5 cm

2.5 cm

Figure 12.15 (a) Periodic lines drawn with one hand when with the othernothing was done or lines and circles were drawn concurrently; (b) periodiccircles drawn with one hand when with the other hand nothing was done orcircles and lines were drawn concurrently (after Franz et al., 1991).

Spijkers and Heuer (1995; Heuer, Spijkers, Kleinsorge, &Steglich, 2000) studied conditions in which one hand had toproduce constant-amplitude oscillations and the other handoscillations with alternating short and long amplitudes. Theyfound that the requirement to change the amplitude in the onehand produced a contralateral effect in addition to the one ob-served with constant-amplitude oscillations. Specifically,after a change from a short to a long amplitude, the amplitudeof the contralateral hand was larger than when only long am-plitudes were repeated, and after a change from a long to ashort amplitude the amplitude of the contralateral hand wassmaller than when only short amplitudes were repeated.These findings indicate that cross-manual effects result notonly from concurrent execution of different amplitudes, butperhaps also from cross-talk between processes of amplitudespecifications. This is also indicated by the observationthat contralateral involuntary amplitude modulations can be

produced not only by movements of alternating short andlong amplitudes, but also by the imagery of such movements(Heuer, Spijkers, Kleinsorge, & van der Loo, 1998).

Discrete-movement studies give clear evidence of a tighttemporal coupling in that movements of different amplitudeand accuracy requirements tend to be of (almost) the same du-ration (Kelso et al., 1979; Marteniuk, MacKenzie, & Baba,1984). However, amplitude assimilation is only weak andtends to be asymmetrical in that the amplitude of a shortermovement is increased, while the amplitude of the concurrentlonger movement is only slightly or not at all reduced(Heuer, Spijkers, Kleinsorge, van der Loo, & Steglich, 1998;Marteniuk et al., 1984; Sherwood, 1991). The tighter temporalthanspatial coupling is also indicatedby the typicalfinding thatmovement durations are more strongly correlated across trialsthan movement amplitudes (e.g., Sherwood, 1991).

Structural constraints on coordination are double-faced:On the one hand, they support the performance of certaintasks, like the production of strictly symmetric movements ofthe upper limbs, and on the other hand they impede the per-formance of other tasks, like the production of asymmetricmovements. Together with the basically soft nature of struc-tural constraints, this suggests that their strength might bemodulated depending on task requirements. In particular,their strength might be enhanced when this is appropriate forthe task at hand, and it might be reduced when this is appro-priate. Such a task-related modulation of structural con-straints on coordination does indeed exist.

Sherwood (1991) found higher intermanual correlationsbetween the amplitudes of discrete rapid reversal movementswhen same amplitudes rather than different amplitudes wererequired. The same result was reported by Heuer, Spijkers,Kleinsorge, van der Loo, and Steglich (1998), who in addi-tion found aftereffects such that the intermanual amplitudecorrelation was higher subsequent to same-amplitude move-ments than following different-amplitude movements, and bySteglich, Heuer, Spijkers, and Kleinsorge (1999) for peakforces of isometric contractions with same and different tar-get forces for the two hands.

Rinkenauer, Ulrich, and Wing (2001) showed for isomet-ric contractions that the requirement to produce differentpeak forces is associated not only with a reduction of the in-termanual correlation between peak forces, but also with a re-duction of the intermanual correlation between rise times.Similarly, when different rise times were to be produced inanother experiment, both the correlations between rise timesand between amplitudes were reduced. These findings indi-cate that the decoupling, which can be observed when differ-ent movements or isometric contractions are to be produced

340 Motor Control

with the two hands, is a generalized phenomenon which isnot restricted to the movement characteristic, that is actuallydifferent. In addition, coupling with respect to peak forcescan be more easily modulated than temporal coupling; infact, peak forces can be fully decoupled, but timing cannot.

In right-handed people, the right and left hands takedifferent roles in bimanual actions: The typical function as-signed to the left hand is holding, while the right hand per-forms manipulations relative to the left one. Generalizing thistypical assignment of functions, Guiard (1987) characterizedthe left and right hand as macrometric and micrometric,respectively. Whereas the left hand is specialized for large-amplitude and low-frequency movements, the right hand isspecialized for accurate small-amplitude and high-frequencymovements. This characterization of the two hands, togetherwith the typical functions assigned to them in bimanual tasks,suggests that structural constraints on coordination may beasymmetric. Although the results on lateral asymmetries inbimanual tasks tend to be somewhat unreliable, there seem tobe at least two consistent findings.

The first finding is that bimanual tasks are easier whenlower-frequency movements are assigned to the left hand andhigher-frequency movements to the right hand than with theopposite assignment of movements to hands. This is true foroscillatory movements (Gunkel, 1962) and discrete fingertaps (Ibbotson & Morton, 1981; Peters, 1981). An example istapping a steady beat with the left hand and a certain rhythmwith the right hand. With the opposite assignment the task isharder, and when performance breaks down, it is typically inthe right hand, which also starts to produce the rhythm as-signed to the left hand. Thus, conforming to Guiard’s (1987)notion of macrometric and micrometric functional specializa-tions of the two hands, task assignments that conform to thesespecializations are easier than task assignments that violatethem. Also conforming to Guiard’s notion, Spijkers andHeuer (1995) observed stronger assimilations of movementamplitudes when large-amplitude oscillations were producedwith the right hand and small-amplitude oscillations with theleft hand than with the opposite assignment of amplitudes tohands.

A second rather consistent finding is a lead of the righthand in bimanual tasks like circle drawing (Stucchi &Viviani, 1993; Swinnen, Jardin, & Meulenbroek, 1996).Stucchi and Viviani (1993) hypothesized that in particular thetiming of bimanual movements might originate from thehemisphere contralateral to the dominant hand. This hypoth-esis has received some support from a PET study (Viviani,Perani, Grassi, Bettinardi, & Fazio, 1998), and it is consistentwith the evidence for tight temporal coupling (or unitarytiming mechanisms) in bimanual tasks.

Levels of Coupling

Structural constraints on coordination can largely be under-stood as resulting from cross-talk between signals involvedin motor control, specifically as a product of coupling, so thatmovements become more similar than intended. Formally,coupling terms are basic ingredients of dynamic models likethe well-known model of Haken et al. (1985). However, thesemodels collapse different kinds of coupling that may exist atdifferent levels of motor control. Such levels can be distin-guished both in functional and in anatomical terms. Here Ishall focus on functionally defined levels. However, it maybe worth mentioning that in anatomical terms there is someevidence for different origins of temporal and spatial cou-pling. For example, split-brain patients give no indication ofa relaxed temporal coupling; if anything, temporal couplingbecomes tighter (Preilowski, 1972; Tuller & Kelso, 1989). Incontrast, spatial coupling seems to be relaxed in split-brainpatients (Franz, Eliassen, Ivry, & Gazzaniga, 1996).

In functional terms, Marteniuk and MacKenzie (1980;Marteniuk et al., 1984) suggested a distinction between anexecution level and a programming level, with cross-talk ef-fects originating at both levels. There can be little doubtabout the existence of cross-talk at the execution level. Thiskind of cross-talk reveals itself in the form of associated ormirror movements, that is, involuntary movements that ac-company voluntary movements of the other hand. These canbe observed in healthy adults (e.g., Durwen & Herzog, 1989,1992; Todor & Lazarus, 1986), but they tend to be more con-spicuous under a variety of neurological conditions or in chil-dren when inhibitory mechanisms, which serve to focus thebasic bilateral innervation unilaterally, are impaired or notyet fully developed (McDowell & Wolff, 1997; Schott &Wyke, 1981). Models that rely on cross-talk at the executionlevel can account for several observations on, for example,bimanual circle drawing (Cattaert, Semjen, & Summers,1999). Nevertheless, there are some results that stronglyfavor the notion of coupling during motor programming (orparametric cross-talk) in addition to cross-talk at the execu-tion level.

Figure 12.16 shows some data obtained with the timed-response procedure (Heuer, Spijkers, Kleinsorge, van derLoo, & Steglich, 1998). The task of the participants was toproduce bimanual reversal movements with short or long am-plitudes. Movements were to be initiated in synchrony withthe last of four pacing tones, and cues were presented at vari-able cuing intervals before the last tone. The cues indicatedthe amplitudes of the movements, which could be short-short, long-long, short-long, and long-short. Participants hadbeen instructed to prepare for movements with intermediate

Motor Coordination 341

Figure 12.16 (a) Mean amplitudes of short and long reversal movements as a function of the cuing interval in a timed-response experiment, shown separatelyfor the left and right hand and the target amplitude of the other hand. (b) Correlations between amplitudes of bimanual reversal movements for the four differentamplitude combinations (after Heuer, Spijkers, Kleinsorge, van der Loo, & Steglich, 1998).

amplitudes and then to produce amplitudes according to thecues as far as this was possible.

The continuous lines in Figure 12.16a show how, with in-creasing preparation time, short and long movements reachtheir final amplitudes, beginning at an intermediate defaultamplitude. Broken lines show the temporal evolution ofthe short and long movement amplitudes when the target am-plitude for the other hand was different, long and short, re-spectively. With long cuing intervals the already describedasymmetric amplitude assimilation was found in that the am-plitude of the short movement was enhanced by the concur-rent requirement to produce a long-amplitude movement

with the other hand, whereas the long-amplitude movementwas not affected by the concurrent short-amplitude move-ment of the other hand. However, at short intervals there wasalso a transient reduction of the long amplitude, although theamplitude difference of the two hands was actually smallerthan at long cuing intervals. In Figure 12.16b the correlationsbetween the amplitudes of the two hands are shown: Theystay at a high level for same-amplitude movements as thecuing interval increases, but they are rapidly reduced for dif-ferent-amplitude movements. These basic findings have beenreplicated for different amplitude differences of the twohands (Heuer, Kleinsorge, Spijkers, & Steglich, 2001) and

342 Motor Control

also for isometric contractions with same and different forces(Steglich et al., 1999).

The data of Figure 12.16a reflect the gradual specificationof movement amplitudes, and they reveal a transient para-metric coupling, that is, a coupling that is gradually relaxedwhen required by the task. This is also indicated by the inter-manual correlations shown in Figure 12.16b. Parametric cou-pling does apply to concurrent specifications of movementparameters, but not to the time-varying force signals or otherexecution-related signals during actual performance of themovements. Thus it should also show up in reaction time,that is, before bimanual movements are actually initiated, andit should show up even in unimanual tasks, provided thatparametric specifications for the movement executed havetemporal overlap with parametric specifications for a move-ment with the other hand that is not produced concurrently.There is indeed some evidence for such effects with move-ments of the two hands with same and different amplitudes(Spijkers, Heuer, Kleinsorge, & van der Loo, 1997; Spijkers,Heuer, Kleinsorge, & Steglich, 2000).

Although parametric coupling seems to be transient as faras amplitude and peak-force specifications are concerned,this is different for temporal specifications. As reviewedabove, for different target durations correlations betweenmovement durations do not decline as strongly as correla-tions between peak forces do. Thus, there is a stronger staticcomponent to the parametric coupling, which accounts forthe fact that it is extremely hard or even impossible toproduce different temporal patterns with the two hands con-currently. If this is indeed the case, one would expect thatreaction time for the choice between a left-hand and a right-hand movement is longer when movements with differentrather than same temporal characteristics are assigned to thetwo hands. The reason is that same temporal characteristicscan be prepared concurrently in advance of the response sig-nal (or perhaps immediately after presentation as long as thechoice of the correct response is not yet finished), whereasthis is impossible for different temporal characteristics. Suchreaction-time differences do exist (see Heuer, 1990, for a re-view; Heuer, 1995).

FLEXIBILITY OF MOTOR CONTROL

The motor transformation, the relation between motor com-mands and resulting movements, is variable. On a short timescale, variations arise when we handle objects, tools, and ma-chines, and when we move in different directions relative togravity. On a longer time scale, variations result from bodygrowth and other bodily changes. As a consequence of such

variations, the internal model of the motor transformation,which captures the relations between motor commands, pro-prioceptive information, and visual information, must beflexible. To study this flexibility experimentally, the relationscan be modified by way of transforming the normal visualinput; in addition, they can be modified by way of adding ex-ternal forces. I shall consider the flexibility of motor controlin both respects in turn.

Adapting and Adjusting to NewVisuo-Motor Transformations

Various kinds of optical transformations can be used tochange the usual relation between movements (motor com-mands and proprioceptive movement information) and theirvisual effects (cf. Welch, 1978, for an overview). The historyof such research dates back to the late nineteenth century,when Stratton (1896, 1897a, 1897b) used spectacles thatserved to turn the visual world upside down. Later Kohler(e.g., 1964) pursued this line of research with various sorts ofdistorting spectacles. All in all, the perceptual consequencesof such severe transformations of the visual world are ex-tremely complex and difficult, if not impossible, to under-stand. However, as far as motor behavior is concerned, thisgenerally comes to appear fairly normal, even if adaptationcan take several days.

A somewhat different and simpler type of transformationof the visual world was introduced by Helmholtz (1867),namely the use of wedge prisms, which serve to shift the vi-sual world laterally. When no visual background (or only ahomogeneous one) is available, the distorting effects ofwedge prisms can be neglected and the consequence of theshifted egocentric visual direction can be studied. A typicallateral displacement is 11°. Thus, when participants are in-structed to point to a target that is visually displaced to theright, their movements will end to the right of the physicaltarget. When they receive feedback on the pointing errors,these will gradually disappear in the course of a series ofmovements. In principle the disappearance of the systematicpointing errors could be due to strategic corrections, that is,to simply pointing to the left of the perceived target. Alterna-tively, it can be due to a change of the internal model of thevisuo-motor transformation. More revealing than the disap-pearance of the pointing error in the exposure phase is thenegative aftereffect that can be observed after removal of thewedge prisms. Now, without visual feedback, subjects tend topoint in the opposite direction, that is, to the left of the targetwhen it had been visually displaced to the right. Negativeaftereffects can also be observed when the prism strength isgradually increased in the exposure period with concurrent

Flexibility of Motor Control 343

displacements of the physical target so that it remains in aconstant egocentric direction; with this procedure, systematicpointing errors are masked by random errors and are notnoticed by the subjects (Howard, 1968).

Prism adaptation implies some kind of change of the in-ternal model of the motor transformation. What is the natureof this change? A first point to note is that it is generalizedand not restricted to the particular movement performed dur-ing prism exposure. Instead, an exposure period with a singlevisual target results in negative aftereffects not only for thisparticular target, but for a range of targets in different direc-tions, and when prismatic displacement is different for tar-gets in different directions, aftereffects reveal a kind of linearinterpolation for targets in between (Bedford, 1989). Beyondthe generalization across the work space, the aftereffect is noteven restricted to pointing at visual targets. In many cases itis approximately the sum of two components, a propriocep-tive aftereffect and a visual aftereffect (Hay & Pick, 1966),with the relative size of the two components depending onexposure conditions (e.g., Kelso, Cook, Olson, & Epstein,1975). A test of the proprioceptive aftereffect is pointingstraight ahead, whereas a test for the visual aftereffect is toalign a visual stimulus with the straight-ahead position. Thus,what is changed is not the specific relation between visualand proprioceptive direction, but rather the directional mean-ing of visual and/or proprioceptive signals.

Some findings suggest that the adaptation during prismexposure does not involve a modification of a single internalmodel of the motor transformation, but some kind of addi-tion, so that the original model remains in existence. On theone hand, negative aftereffects decay even when participantsremain in darkness (e.g., Dewar, 1971); on the other hand,even when participants are confronted with the normal visualworld between experimental sessions, long-term effects ofprismatic adaptation can be observed as soon as they arebrought back to the experimental setup (McGonigle & Flook,1978). In addition, with repeated alternations of periods withand without lateral displacement, or with different lateral dis-placements, aftereffects tend to disappear, and switching be-tween different visuo-motor transformations becomes almostinstantaneous (e.g., Kravitz, 1972; Welch, 1971). These andother results (cf. Welch, Bridgeman, Anand, & Browman,1993) strongly suggest that multiple models of visuo-motortransformations can be learned and, when required bythe task, selectively be put to use, although little is knownabout the nature of the cues that mediate the retrieval ofstored internal models.

In a certain way, the situation when wearing laterally dis-placing prisms is similar to a situation that has become quitecommon during the last one or two decades, namely the

operation of a computer mouse with concurrent movementsof a cursor on a laterally displaced monitor. Although in thefirst case there are aftereffects, we encounter no difficulties inoperating the mouse with different lateral displacements ofthe screen. There seem to be mainly two reasons for this dif-ference: First, movements performed with the computermouse are parameterized in terms of (allocentric) distances,whereas movements produced in experiments with laterallydisplacing prisms are parameterized in terms of (egocentric)locations. Second, and perhaps of less importance, is thatwith the computer mouse proprioceptive and visual informa-tion indicate different egocentric directions of differentobjects, the hand and the cursor, while in prism-adaptationstudies they refer to the same object, the hand. Object iden-tity is a factor that affects the size of negative aftereffects(Welch, 1972).

Visuo-motor transformations can also be changed suchthat the relations between (allocentric) visual distancesand/or directions and (hand-centered) movement amplitudesand/or directions are modified. The basic findings seem toparallel those obtained in prism-adaptation studies to a re-markable degree. For example, aftereffects occur and multi-ple models of visuo-motor transformations can be learnedand selectively accessed when appropriate (Cunningham &Welch, 1994). When a certain internal model has beenlearned, there seems to follow a kind of labile period in whichthe learning of a new transformation results in a modificationof the model, but after a period of consolidation the learningof a new transformation results in the development of a newmodel rather than the overriding of the old one (Krakauer,Ghilardi, & Ghez, 1999). With sufficient delays betweenlearning periods, it seems that repeated alternation betweendifferent transformations is not needed to acquire multiple in-ternal models.

When discussing visual feedback, I have pointed to thelimitations in acquiring internal models of additional trans-formations of one’s own movements. Whereas adjustments tochanges of visuo-motor gains require only one or a few dis-crete movements (Young, 1969), adjustments to new rela-tions between the directions of hand movements and cursormotions require more trials (e.g., Krakauer et al., 1999). Ad-justments to nonlinear transformations require even longerexperience, and for too complex transformations internalmodels can no longer be developed. Although the mastery ofadded transformations is typically not associated with theirawareness (who could tell the gain factor of his or her com-puter mouse?), the difficulty of such transformations isaffected by higher level cognitive processes. This is nicelyillustrated by a little-known study of Merz, Kalveram, andHuber (1981), and additional evidence from reaction-time

344 Motor Control

Figure 12.17 Lifting an object. Period a is the preload phase, b the loadingphase, c the transitional phase, in which the object is actually lifted, followedby the static or hold phase (after Johansson, 1996).

studies is reported in the chapter by Proctor and Vu in thisvolume.

In the experiment of Merz et al. (1981), participants con-trolled the movement of a cursor on an oscilloscope screen bymeans of lateral pressure on a knob. Their task was a trackingtask in which they had to keep the cursor aligned with a mov-ing target. For two groups of participants the cursor moved tothe right when the knob was pressed to the right (compatiblerelation), and for two groups of participants the relation wasreversed (incompatible). One group with each of the two lev-els of compatibility had the knob placed at the bottom of asteering wheel, while for the other two groups the steeringwheel was covered by a piece of cardboard. In the incompat-ible conditions there was a strong practice effect. Moreimportant, however, is that the performance deficit in the in-compatible conditions disappeared when the steering wheelwas visible; performance in this condition was as good as inthe compatible conditions. Thus, the visibility of the steeringwheel enabled the subjects to change the incompatible rela-tion to a compatible one in associating clockwise rotation ofthe wheel, which is consequent upon leftward pressure, witha rightward motion of the cursor on the screen.

Adjusting and Adapting to External Forces

When external forces vary, motor commands for intendedmovements have to be modulated accordingly. Except formovements with different directions relative to gravity, per-haps the most frequent adjustments are required when wedeal with objects of different masses. Here, depending on themass, the kinematic characteristics vary. Specifically, withincreasing mass, peak acceleration and peak deceleration aswell as peak velocity tend to decline, while movement dura-tion tends to increase. This is true both for lifting objects(Gachoud, Mounoud, Hauert, & Viviani, 1983) and for mov-ing them in a horizontal plane (Gottlieb, Corcos, & Agarwal,1989). Thus, for objects with different masses, peak forcesare not perfectly scaled, which would result in an invariantacceleration profile; instead, with increasing mass, accelera-tion and deceleration become smaller, but an increasing dura-tion serves to avoid a shortening of the amplitude of themovements. In spite of these mass-dependent variations, thebasic shape of the velocity profile remains invariant; that is,with the proper scaling of time and velocity, the profiles be-come identical (Bock, 1990; Ruitenbeek, 1984).

In lifting an object, it is not only the so-called load forcewhich has to be adjusted to the mass (and weight) of the ob-ject, but also the grip force (cf. Johansson, 1996, for review).When the grip force is too weak, the object may slip; when itis too strong, the object may break; and, in addition, too high

forces are uneconomical. Figure 12.17 shows the buildup ofboth types of force when an object is lifted. First, grip forcestarts to develop (preload phase), then load force (loadingphase). When the load force is sufficiently strong, the objectis lifted (transitional phase) and thereafter held in a certainposition.

During the loading phase a certain relation between gripforce and load force is established, which is generally some-what higher than the minimal value required to preventthe object from slipping; this safety margin is typically in therange of 10–40%. Of course, the proper adjustment of thegrip force depends not only on the object, but also on its sur-face characteristics. In fact, there is a delicate grip-force ad-justment to the friction between fingers and object surfacethat depends not only on the surface characteristics of the ob-ject, but also on those of the skin, which change, for example,after washing one’s hands.

Adjustment of grip force is required not only when an ob-ject is lifted, but also when it is moved around, so that thereis an inertial load in addition to the gravitational load. In amanner similar to the way load force and grip force increasein parallel when an object is lifted, grip force is modulated inparallel to inertial load while an object is moved (Flanagan,Tresilian, & Wing, 1993; Flanagan & Wing, 1995). In mov-ing an object, there is an important difference betweenperiodic horizontal and vertical movements. In horizontalmovements, inertial load is orthogonal to gravitational load;inertial load reaches maxima both at the left and right move-ment reversals, and grip force reaches maxima at these pointsas well. Thus, there is a 1:2 ratio of the frequencies of peri-odic movements and grip-force modulations. In contrast, forvertical movements inertial and gravitational load add, sothat total load is particularly strong at the lower movement

Flexibility of Motor Control 345

reversal, but it can be close to zero at the upper movement re-versal. In this case the frequency ratio becomes 1:1. This dif-ference between horizontal and vertical movements disap-pears under conditions of microgravity because of theabsence of gravitational load, and subjects adapt rapidly (dur-ing parabolic flights) to produce the appropriate 1:2 ratioalso with vertical movements (Hermsdörfer et al., 2000).

Force adjustments are both predictive and reactive. Thisbasic principle is evident already from the everyday observa-tion that when someone hands an object to someone else, thelatter can normally hold it without difficulties; only when theobject is unexpectedly light or heavy, short-lived difficultiesarise. Without knowing about the change of the mass of amoved object, the first movement is perturbed, but the nextmovement is properly adjusted to the new load (Bock, 1993).In fact, corrections for the unexpected mass set in as early asduring the first movement, which, in the case of an unexpect-edly high load, results mainly in a prolonged movement du-ration (Bock, 1993; Smeets, Erkelens, & Denier van der Gon,1995). It seems that under conditions of microgravity, whenobjects are weightless but nevertheless have normal mass andthus inertial load, movements exhibit characteristics ofmovements with an unexpectedly high mass even for weeks(Sangals, Heuer, Manzey, & Lorenz, 1999).

Motor commands for active movements are most likelyamong the information that is involved in predictive force ad-justments, as can be evidenced from the grip-force adjust-ments while moving a hand-held object. Of course, this kindof prediction can work only when force adjustments are re-quired as a consequence of self-generated activity. Whenforce adjustments are required to accommodate variations inload, which are independent of self-generated activity, pre-dictions must rely on other kinds of information. Obviously,proper force adjustments depend on experience; the firstmovement after an unnoticed change of the load is performedwith an initially maladjusted force, but not the second one. Inaddition, seen object size plays a role, although force adjust-ment is not necessarily related to estimates of weight.Gordon, Forssberg, Johansson, and Westling (1991) exam-ined the lifting of boxes of identical weight but differentsizes. Although subjects judged the smaller boxes to be heav-ier than the larger ones, peak grip and load forces werestronger for the larger ones. However, this difference waspresent only during lifting and disappeared during subse-quent holding of the object, when force adjustments wereperhaps related to the actual weight and no longer to the vi-sually mediated predictive mechanisms.

Adjustments to objects of different masses seem to becomparatively simple achievements, similar to adjustmentsto different visuo-motor gains. When the external forces

which act on a moving limb are transformed in a morecomplex way, similarities between adjustments to modifiedvisuo-motor transformations and modified external forcefields become more conspicuous. Shadmehr and Mussa-Ivaldi (1994) introduced such forces while the participantsmoved a robot arm. For example, forces were proportional tohand velocity and nearly orthogonal to the direction of handmovement, so that initially the paths of the hand werestrongly curved. With continued experience the paths becameagain approximately straight lines, which—as an aside—canbe taken as additional evidence for the claim that motor plan-ning refers to end-effector kinematics. After removal of theexternal forces there were negative aftereffects: The paths ofthe hand were curved again, but in the opposite direction. Theaftereffects indicate that the adjustments were based on a newinternal model of the dynamic transformation.

Similar to the findings with modified visuo-motor trans-formations, multiple internal models of dynamic transforma-tions can be acquired (Shadmehr & Brashers-Krug, 1997).Again there seems to be a labile period after a new model hasbeen learned, during which it will be unlearned when anotherdynamic transformation is experienced, but after a period ofconsolidation this is no longer the case. Once an internalmodel has survived the labile period, it can be put to efficientuse even months later.

Adjustments to new dynamic transformations generalizeacross different types of movement (Condit, Gandolfo, &Mussa-Ivaldi, 1997) and also across the work space(Shadmehr & Mussa-Ivaldi, 1994). When movements areperformed in a different region of the workspace from duringthe practice period, external forces can remain invarianteither with respect to the movement of the end-effector incoordinates of extrinsic space or with respect to the jointmovements. Generalizations across the work space turnedout to be approximate in joint coordinates. This is consistentwith a particularly intriguing parallel between adaptation toshifted visual directions and additional external forces.

Prism adaptation involves modified relations between pro-prioceptive and/or visual signals and their meaning in termsof egocentric directions. Adaptation to a modified externalforce field seems to involve a modified relation between mus-cle activations (or motor commands) and the directions ofconsequent movements (Shadmehr & Moussavi, 2000). Forexample, the EMG signal from an elbow flexor can be plottedas a function of movement direction (more precisely, it is theEMG signal integrated across a certain time interval aroundthe start of the movement); this results in a directional tuningcurve of a muscle. Of course, the peak of this curve is shiftedwhen the shoulder joint is moved. However, there is also ashift induced by the adaptation to a new force field, and this

346 Motor Control

shift is more or less additive to shifts associated with rota-tions at the shoulder. This adaptation-induced shift allowsone to predict the generalization across the work space.

MOVING ON

Research on the adaptive capabilities of human motor controlcan serve to illustrate some fairly general characteristics ofthe field: First, different phases or waves can be distin-guished; second, different lines of research come togetherand trigger new waves; third, research on applied problemsoften precedes more theoretically minded research; andfourth, new concepts enter the field, often coming from otheracademic disciplines. At present, research on adapting to vi-sual distortions and to added transformations, as in trackingtasks, is combined; research on the latter has a strong appliedhistory related, for example, to vehicle control. The new the-oretical concepts come largely from modern control theoryand robotics. On top of such developments are new measure-ment technologies, which have made the recording of move-ments easier and which open progressively wider windowsonto the activity of the brain while it controls movement.

Science seems to be driven largely by practical needs andby the apparent human desire to have coherent ideas of one-self and the world one lives in. Perhaps some of the findingsreported in this chapter challenge ideas humans tend to haveabout themselves, but as far as motor control is concerned, themore important driving forces seem to be practical, related totechnical developments as in robotics, to the control of com-plex machines, and to new challenges for manual skills, as inminimally invasive surgery. Perhaps future developmentswill result in tighter links of the (functional) theoretical con-cepts of the field to the solution of applied problems on theone hand and to the neuronal substrates on the other hand.

REFERENCES

Abbs, J. H., Gracco, V. L., & Cole, K. J. (1984). Control of multi-movement coordination: Sensorimotor mechanisms in speechmotor programming. Journal of Motor Behavior, 16, 195–232.

Abrams, R. A., & Landgraf, J. Z. (1990). Differential use of distanceand location information for spatial localization. Perception &Psychophysics, 47, 349–359.

Aschersleben, G., Gehrke, J., & Prinz, W. (2001). Tapping withperipheral nerve block. A role for tactile feedback in the timingof movements. Experimental Brain Research, 136, 331–339.

Aschersleben, G., & Prinz, W. (1995). Synchronizing actions withevents: The role of sensory information. Perception & Psy-chophysics, 57, 305–317.

Aschersleben, G., & Prinz, W. (1997). Delayed auditory feedback insynchronization. Journal of Motor Behavior, 29, 35–46.

Baldissera, F., Cavallari, P., & Civaschi, P. (1982). Preferentialcoupling between voluntary movements of ipsilateral limbs.Neuroscience Letters, 34, 95–100.

Baldissera, F., Cavallari, P., Marini, G., & Tassone, G. (1991).Differential control of in-phase and anti-phase coupling of rhyth-mic movements of ipsilateral hand and foot. Experimental BrainResearch, 83, 375–380.

Bedford, F. L. (1989). Constraints on learning new mappings be-tween perceptual dimensions. Journal of Experimental Psychol-ogy: Human Perception and Performance, 15, 232–248.

Beek, P. J. (1989). Juggling dynamics. Amsterdam: Free UniversityPress.

Bennett, S., van der Kamp, J., Savelsbergh, G. J., & Davids, K.(1999). Timing a one-handed catch. I. Effects of telestereoscopicviewing. Experimental Brain Research, 129, 362–368.

Bizzi, E., Accornero, N., Chapple, W., & Hogan, N. (1984). Posturecontrol and trajectory formation during arm movement. Journalof Neuroscience, 4, 2738–2744.

Blöte, A. W., & Dijkstra, J. F. (1989). Task effects on young chil-dren’s performance in manipulating a pencil. Human MovementScience, 8, 515–528.

Bock, O. (1990). Load compensation in human goal-directed move-ments. Behavioural Brain Research, 41, 167–177.

Bock, O. (1993). Early stages of load compensation in human aimedarm movements. Behavioural Brain Research, 55, 61–68.

Bock, O., & Arnold, K. (1993). Error accumulation and error cor-rection in sequential pointing movements. Experimental BrainResearch, 95, 111–117.

Bock, O., & Eckmiller, R. (1986). Goal-directed arm movements inabsence of visual guidance: Evidence for amplitude rather thanposition control. Experimental Brain Research, 62, 451–458.

Bootsma, R. J. (1989). Accuracy of perceptual processes subservingdifferent perception-action systems. Quarterly Journal of Exper-imental Psychology, 41A, 489–500.

Bootsma, R. J., & van Wieringen, P. C. W. (1990). Timing an attack-ing forehand drive in table tennis. Journal of Experimental Psy-chology: Human Perception and Performance, 16, 21–29.

Bridgeman, B., Kirch, M., & Sperling, A. (1981). Segregation ofcognitive and motor aspects of visual function using inducedmotion. Perception & Psychophysics, 29, 336–342.

Bridgeman, B., Lewis, S., Heit, G., & Nagle, M. (1979). Relationbetween cognitive and motor-oriented systems of visual positionperception. Journal of Experimental Psychology: Human Per-ception and Performance, 5, 692–700.

Bridgeman, B., Peery, S., & Anand, S. (1997). Interaction of cogni-tive and sensorimotor maps of visual space. Perception & Psy-chophysics, 59, 456–469.

Brown, T. G. (1911). The intrinsic factors in the act of progressionin the mammal. Proceedings of the Royal Society, 84B, 308–319.

References 347

Brunia, C. H. M. (1999). Neural aspects of anticipatory behavior.Acta Psychologica, 101, 213–242.

Brunia, C. H. M., Haagh, S. A. V. M., & Scheirs, J. G. M. (1985).Waiting to respond: Electrophysiological measurements in manduring preparation for a voluntary movement. In H. Heuer,U. Kleinbeck, & K.-H. Schmidt (Eds.), Motor behavior: Pro-gramming, control, and acquisition (pp. 35–78). Berlin, Ger-many: Springer.

Bullock, D., & Grossberg, S. (1988). Neural dynamics of planned armmovements: Emergent invariants and speed-accuracy propertiesduring trajectory formation. Psychological Review, 95, 49–90.

Campion, J., Latto, R., & Smith, Y. M. (1983). Is blindsight an effectof scattered light, spared cortex, and near-threshold vision?Behavioral and Brain Sciences, 6, 423–486.

Carlton, L. G. (1981). Visual information: The control of aimingmovements. Quarterly Journal of Experimental Psychology,33A, 87–93.

Carson, R. G., Thomas, J., Summers, J. J., Walters, M. R., & Semjen,A. (1997). The dynamics of bimanual circle drawing. QuarterlyJournal of Experimental Psychology, 50A, 664–683.

Cattaert, D., Semjen, A., & Summers, J. J. (1999). Simulating aneural cross-talk model for between-hand interference duringbimanual circle drawing. Biological Cybernetics, 81, 343–358.

Chang, P., & Hammond, G. R. (1987). Mutual interactions betweenspeech and finger movements. Journal of Motor Behavior, 19,265–274.

Cohen, L. (1971). Synchronous bimanual movements performed byhomologous and nonhomologous muscles. Perceptual and MotorSkills, 32, 639–644.

Condit, M. A., Gandolfo, F., & Mussa-Ivaldi, F. (1997). The motorsystem does not learn the dynamics of the arm by role memo-rizations of past experience. Journal of Neurophysiology, 78,554–560.

Cooke, J. D. (1980). The organization of simple, skilled movements.In G. E. Stelmach & J. Requin (Eds.), Tutorials in motor behav-ior (pp. 199–212). Amsterdam: North-Holland.

Cordo, P. J., & Nashner, L. M. (1982). Properties of postural adjust-ments associated with rapid arm movements. Journal of Neuro-physiology, 47, 287–302.

Crossman, E. R. F. W., & Goodeve, P. J. (1963/1983). Feedback con-trol of hand-movement and Fitts’ law. Quarterly Journal of Ex-perimental Psychology, 35A, 251–278 (originally presented at theMeeting of the Experimental Psychology Society, Oxford, 1963).

Cruse, H. (1986). Constraints for joint angle control of the humanarm. Biological Cybernetics, 54, 125–132.

Cruse, H., & Brüwer, M. (1987). The human arm as a redundantmanipulator: The control of path and joint angles. BiologicalCybernetics, 57, 137–144.

Cruse, H., Dean, J., Heuer, H., & Schmidt, R. A. (1990). Utilizationof sensory information for motor control. In O. Neumann &W. Prinz (Eds.), Relationships between perception and action:Current approaches (pp. 43–79). Berlin, Germany: Springer.

Cunningham, H. A., & Welch, R. B. (1994). Multiple concurrentvisual-motor mapping: implications for models of adaptation.Journal of Experimental Psychology: Human Perception andPerformance, 20, 987–999.

Darling, W. G., Cole, K. J., & Abbs, J. H. (1988). Kinematic vari-ability in grasp movements as a function of practice and move-ment speed. Experimental Brain Research, 73, 225–235.

Davidson, P. R., Jones, R. D., Sirisena, H. R., & Andreae, J. H.(2000). Detection of adaptive inverse models in the humanmotor system. Human Movement Science, 19, 761–795.

Derwort, A. (1938). Untersuchungen über den Zeitablauf figurierterBewegungen beim Menschen [Studies of the timing of figuredmovements in man]. Pflügers Archiv für die gesamte Physiolo-gie, 240, 661–675.

Dewar, R. (1971). Adaptation to displaced vision: Variationsin the shaping technique. Perception & Psychophysics, 9,155–157.

Deutsch, D. (1983). The generation of two isochronous sequences inparallel. Perception & Psychophysics, 34, 331–337.

Durwen, H. F., & Herzog, A. G. (1989). Electromyographic investi-gation of mirror movements in normal adults: Variation of fre-quency with side of movements, handedness, and dominance.Brain Dysfunction, 2, 84–92.

Durwen, H. F., & Herzog, A. G. (1992). Electromyographic investi-gation of mirror movements in normal adults: Variation offrequency with site, effort, and repetition of movement. BrainDysfunction, 5, 310–318.

Elliott, D., & Allard, F. (1985). The utilization of visual feedback in-formation during rapid pointing movements. Quarterly Journalof Experimental Psychology, 37A, 407–425.

Elliott, D., & Lee, T. D. (1995). The role of target information onmanual-aiming bias. Psychological Research, 58, 2–9.

Elliott, D., & Madalena, J. (1987). The influence of premovementvisual information on manual aiming. Quarterly Journal ofExperimental Psychology, 39A, 541–559.

Engström, D. A., Kelso, J. A. S., & Holroyd, T. (1996). Reaction-anticipation transitions in human perception-action patterns.Human Movement Science, 15, 809–832.

Favilla, M., & De Cecco, E. (1996). Parallel direction and extentspecification of planar reaching arm movements in humans.Neuropsychology, 34, 609–613.

Favilla, M., Hening, W., & Ghez, C. (1989). Trajectory control intargeted force impulses. VI. Independent specification of re-sponse amplitude and direction. Experimental Brain Research,75, 280–294.

Fisk, J. D., & Goodale, M. A. (1989). The effects of instructions tosubjects on the programming of visually directed reachingmovements. Journal of Motor Behavior, 21, 5–19.

Fitts, P. M. (1954). The information capacity of the human motorsystem in controlling the amplitude of movement. Journal ofExperimental Psychology, 47, 381–391.

348 Motor Control

Flanagan, J. R., Tresilian, J., & Wing, A. M. (1993). Coupling ofgrip force and load force during arm movements with graspedobjects. Neuroscience Letters, 152, 53–56.

Flanagan, J. R., & Wing, A. M. (1995). The stability of precisiongrip forces during cyclic arm movements with a hand-held load.Experimental Brain Research, 105, 455–464.

Flanders, M., Helms Tillery, S. I., & Soechting, J. F. (1992). Earlystages in a sensorimotor transformation. Behavioral and BrainSciences, 15, 309–362.

Flash, T., & Hogan, N. (1985). The coordination of arm movements:An experimentally confirmed mathematical model. Journal ofNeuroscience, 5, 1688–1703.

Franz, E. A. (1997). Spatial coupling in the coordination of complexactions. Quarterly Journal of Experimental Psychology, 50A,684–704.

Franz, E. A., Eliassen, J. C., Ivry, R. B., & Gazzaniga, M. S. (1996).Dissociation of spatial and temporal coupling in the bimanualmovements of callosotomy patients. Psychological Science, 7,306–310.

Franz, E. A., Zelaznik, H. N., & McCabe, G. (1991). Spatial topo-logical constraints in a bimanual task. Acta Psychologica, 77,137–151.

Gachoud, J. P., Mounoud, P., Hauert, C. A., & Viviani, P. (1983).Motor strategies in lifting movements: A comparison of adultand child performance. Journal of Motor Behavior, 15, 202–216.

Gentilucci, M., Chiefi, S., Daprati, E., Saetti, M. C., & Toni, I. (1996).Visual illusion and action. Neuropsychologia, 34, 369–376.

Gentner, D. R. (1987). Timing of skilled motor performance: Testsof the proportional duration model. Psychological Review, 94,255–276.

Georgopoulos, A. P., Lurito, J. T., Petrides, M., Schwartz, A. B., &Massey, J. T. (1989). Mental rotation of the neuronal populationvector. Science, 243, 234–236.

Georgopoulos, A. P., & Massey, J. T. (1987). Cognitive spatial-motor processes. 1. The making of movements at various anglesfrom a stimulus direction. Experimental Brain Research, 65,361–370.

Ghez, C., Favilla, M., Ghilardi, M. F., Gordon, J., Bermejo, R., &Pullman, S. (1997). Discrete and continuous planning of handmovements and isometric force trajectories. Experimental BrainResearch, 115, 217–233.

Gielen, S. C. A. M., van den Heuvel, P. J. M., & van Gisbergen,J. A. M. (1984). Coordination of fast eye and arm movements ina tracking task. Experimental Brain Research, 56, 154–161.

Gillam, B., & Chambers, D. (1985). Size and position are incongru-ous: Measurements on the Müller-Lyer figure. Perception &Psychophysics, 37, 549–556.

Glencross, D. J. (1970). Serial organization and timing in a motorskill. Journal of Motor Behavior, 2, 229–237.

Goodale, M. A., & Humphrey, G. K. (1998). The objects of actionand perception. Cognition, 67, 181–207.

Goodale, M. A., & Milner, A. D. (1992). Separate visual pathwaysfor perception and action. Trends in Neuroscience, 15, 20–25.

Goodman, D., & Kelso, J. A. S. (1980). Are movements preparatedin parts? Not under compatible (naturalized) conditions. Journalof Experimental Psychology: General, 109, 475–495.

Goodwin, G. M., McCloskey, D. I., & Matthews, P. B. C. (1972).The contribution of muscle afferents to kinaesthesia shown byvibration induced illusions of movement and by the effects ofparalysing joint afferents. Brain, 95, 705–748.

Gordon,A.M.,Forssberg,H., Johansson,R.S.,&Westling,G. (1991).Visual size cues in the programming of manipulative forces duringprecision grip. Experimental Brain Research, 83, 477–482.

Gordon, J., Ghilardi, M. F., & Ghez, C. (1994). Accuracy of planarreaching movements. I. Independence of direction and extentvariability. Experimental Brain Research, 99, 97–111.

Gottlieb, G. L., Corcos, D. M., & Agarwal, G. C. (1989). Organiz-ing principles for single joint movements. I: A speed-insensitivestrategy. Journal of Neurophysiology, 62, 342–357.

Grossberg, S., Pribe, C., & Cohen, M. A. (1997). Neural control ofinterlimb oscillations: I. Human bimanual coordination. Biolog-ical Cybernetics, 77, 131–140.

Guiard, Y. (1987). Asymmetric division of labor in human skilledbimanual action: The kinematic chain as a model. Journal ofMotor Behavior, 19, 486–517.

Gundry, J. (1975). The use of location and distance in reproducingdifferent amplitudes of movement. Journal of Motor Behavior, 7,91–100.

Gunkel, M. (1962). Über relative Koordination bei willkürlichenmenschlichen Gliederbewegungen [On relative coordination involuntary human limb movements]. Pflügers Archiv für diegesamte Physiologie, 275, 472–477.

Haken, H. (1982). Synergetik: Eine Einführung [Synergetics: Anintroduction]. Berlin, Germany: Springer.

Haken, H., Kelso, J. A. S., & Bunz, H. (1985). A theoretical modelof phase transitions in human hand movements. BiologicalCybernetics, 51, 347–356.

Hay, J. C., & Pick, H. (1966). Visual and proprioceptive adaptationto optical displacement of the visual stimulus. Journal of Exper-imental Psychology, 71, 150–158.

Helmholtz, H. von. (1867). Handbuch der Physiologischen Optik[Handbook of Physiological Optics]. Leipzig, Germany: Voss.

Hening, W., Favilla, M., & Ghez, C. (1988). Trajectory control intargeted force impulses. V. Gradual specification of response am-plitude. Experimental Brain Research, 71, 116–128.

Henry, F. M., & Rogers, D. E. (1960). Increased response latency forcomplicated movements and a “memory drum” theory of neuro-motor reaction. Research Quarterly, 31, 448–458.

Hermsdörfer, J., Marquardt, C., Philipp, J., Zierdt, A., Nowak, D.,Glasauer, S., & Mai, N. (2000). Moving weightless objects. Gripforce control during microgravity. Experimental Brain Research,132, 52–64.

References 349

Heuer, H. (1983). Bewegungslernen [Motor learning]. Stuttgart,Germany: Kohlhammer.

Heuer, H. (1989). Movement strategies as points on equal-outcomecurves. Behavioral and Brain Sciences, 12, 220–221.

Heuer, H. (1990). Rapid responses with the left or right hand:Response-response compatibility effects due to intermanualinteractions. In R. W. Proctor & T. G. Reeve (Eds.), Stimulus-response compatibility: An integrated perspective (pp. 311–342). Amsterdam: North-Holland.

Heuer, H. (1991). Invariant relative timing in motor-program theory.In J. Fagard & P. H. Wolff (Eds.), The development of timingcontrol and temporal organization in coordinated action: Invari-ant relative timing, rhythms, and coordination (pp. 37–68).Amsterdam: North-Holland.

Heuer, H. (1993a). Estimates of time to contact based on changingsize and changing target vergence. Perception, 22, 549–563.

Heuer, H. (1993b). Structural constraints on bimanual movements.Psychological Research, 55, 83–98.

Heuer, H. (1995). Models for response-response compatibility: Theeffects of the relation between responses in a choice task. ActaPsychologica, 90, 315–332.

Heuer, H. (1998). Blocking in rapid finger tapping: The role ofvariability in proximo-distal coordination. Journal of MotorBehavior, 30, 130–142.

Heuer, H., Kleinsorge, T., Spijkers, W., & Steglich, C. (2001). Staticand phasic cross-talk effects in discrete bimanual reversal move-ments. Journal of Motor Behavior, 33, 67–85.

Heuer, H., & Sangals, J. (1998). Task-dependent mixtures of coor-dinate systems in visuomotor transformations. ExperimentalBrain Research, 119, 224–236.

Heuer, H., & Schmidt, R. A. (1988). Transfer of learning amongmotor patterns with different relative timing. Journal of Experi-mental Psychology: Human Perception and Performance, 14,241–252.

Heuer, H., Schmidt, R. A., & Ghodsian, D. (1995). Generalizedmotor programs for rapid bimanual tasks: A two-levelmultiplicative-rate model. Biological Cybernetics, 73, 343–356.

Heuer, H., Spijkers, W., Kleinsorge, T., & Steglich, C. (2000).Parametrische Kopplung bei Folgen beidhändiger Umkehrbewe-gungen mit gleichen und unterschiedlichen Weiten [Parametriccoupling in sequences of bimanual reversal movements withsame and different amplitudes]. Zeitschrift für experimentellePsychologie, 47, 34–49.

Heuer, H., Spijkers, W., Kleinsorge, T., & van der Loo, H. (1998a).Period duration of physical and imaginary movement sequencesaffects contralateral amplitude modulation. Quarterly Journal ofExperimental Psychology, 51A, 755–779.

Heuer, H., Spijkers, W., Kleinsorge, T., van der Loo, H., & Steglich,C. (1998b). The time course of cross-talk during the simultane-ous specification of bimanual movement amplitudes. Experi-mental Brain Research, 118, 381–392.

Hollerbach, J. M., & Atkeson, C. G. (1987). Deducing planningvariables from experimental arm trajectories: Pitfalls and possi-bilities. Biological Cybernetics, 56, 279–292.

Holst, E. von (1939). Die relative Koordination als Phänomen undals Methode zentralnervöser Funktionsanalyse [Relative coordi-nation as phenomenon and as a method for the functional analy-sis of the central nervous system]. Ergebnisse der Physiologie,42, 228–306.

Howard, I. P. (1968). Displacing the optical array. In S. J. Freedman(Ed.), The neuropsychology of spatially oriented behavior(pp. 19–36). Homewood, IL: Dorsey Press.

Ibbotson, N. R., & Morton, J. (1981). Rhythm and dominance. Cog-nition, 9, 125–138.

Jagacinski, R. J., Marshburn, E., Klapp, S. T., & Jones, M. R.(1988). Tests of parallel versus integrated structure in polyrhyth-mic tapping. Journal of Motor Behavior, 20, 416–442.

James, W. (1950). The principles of psychology. New York: Dover.(Original work published 1890)

Jeannerod, M. (1984). The timing of natural prehension movements.Journal of Motor Behavior, 16, 235–254.

Johansson, R. S. (1996). Sensory and memory information in thecontrol of dexterous manipulation. In F. Lacquaniti & P. Viviani(Eds.), Neural bases of motor behaviour (pp. 205–260).Dordrecht, The Netherlands: Kluwer.

Jordan, M. I. (1996). Computational aspects of motor control andmotor learning. In H. Heuer & S. W. Keele (Eds.), Handbook ofperception and action: Vol. 2. Motor skills (pp. 71–120). London:Academic Press.

Jung, R., & Fach, C. (1984). Spiegelschrift und Umkehrschrift beiLinkshändern und Rechtshändern: Ein Beitrag zum Balkentrans-fer und Umkehrlernen [Mirrorscript and invalid script in left-handers and righthanders: A contribution to colossal transfer andlearned inversion]. In L. Spillmann & B. R. Wooten (Eds.), Sen-sory experience, adaptation, and perception: Festschrift for IvoKohler (pp. 377–399). Hillsdale, NJ: Erlbaum.

Kalveram, K.-T. (1991). Pattern generating and reflex-like pro-cesses controlling aiming movements in the presence of inertia,damping and gravity. Biological Cybernetics, 64, 413–419.

Kay, B. A., Kelso, J. A. S., Saltzman, E., & Schöner, G. (1987).Space-time behavior of single and bimanual rhythmical move-ments: Data and limit cycle model. Journal of Experimental Psy-chology: Human Perception and Performance, 13, 178–192.

Keele, S. W. (1968). Movement control in skilled motor perfor-mance. Psychological Bulletin, 70, 387–403.

Keele, S. W. (1986). Motor control. In K. R. Boff, L. Kaufman, &J. P. Thomas (Eds.), Handbook of perception and performance:Vol. 2. Cognitive processes and performance (pp. 30-1–30-60).New York: Wiley.

Keele, S. W., & Posner, M. I. (1968). Processing of visual feedbackin rapid movements. Journal of Experimental Psychology, 77,155–158.

350 Motor Control

Kelso, J. A. S. (1984). Phase transitions and critical behaviorin human bimanual coordination. American Journal of Physiol-ogy: Regulatory, Integrative, and Comparative, 246, R1000–R1004.

Kelso, J. A. S. (1994). Elementary coordination dynamics. In S. P.Swinnen, H. Heuer, J. Massion, & P. Casaer (Eds.), Interlimbcoordination: Neural, dynamical, and cognitive constraints(pp. 301–318). San Diego, CA: Academic Press.

Kelso, J. A. S., Cook, E., Olson, M. E., & Epstein, W. (1975). Allo-cation of attention and the locus of adaptation to displacedvision. Journal of Experimental Psychology: Human Perceptionand Performance, 1, 237–245.

Kelso, J. A. S., & Holt, K. G. (1980). Exploring a vibratory systemanalysis of human movement production. Journal of Neurophys-iology, 43, 1183–1196.

Kelso, J. A. S., Southard, D. L., & Goodman, D. (1979). On thecoordination of two-handed movements. Journal of Experimen-tal Psychology: Human Perception and Performance, 5, 229–238.

Kelso, J. A. S., Tuller, B., & Harris, K. S. (1983). A “dynamicpattern” perspective on the control and coordination of move-ment. In P. F. MacNeilage (Ed.), The production of speech(pp. 137–173). Berlin, Germany: Springer.

Kelso, J. A. S., Tuller, B., Vatikiotis-Bateson, E., & Fowler, C. A.(1984). Functionally specific articulatory cooperation followingjaw perturbation during speech: Evidence for coordinative struc-tures. Journal of Experimental Psychology: Human Perceptionand Performance, 10, 812–832.

Klapp, S. T. (1979). Doing two things at once: The role of temporalcompatibility. Memory and Cognition, 7, 375–381.

Klapp, S. T. (1981). Temporal compatibility in dual motor tasks II:simultaneous articulation and hand movements. Memory andCognition, 9, 398–401.

Klapp, S. T., Hill, M., Tyler, J., Martin, Z., Jagacinski, R., & Jones,M. (1985). On marching to two different drummers: Perceptualaspects of the difficulties. Journal of Experimental Psychology:Human Perception and Performance, 11, 814–828.

Klemm, O. (1938). Zwölf Leitsätze zu einer Psychologie derLeibesübungen [Twelve guiding principles for a psychology ofgymnastics]. Neue Psychologische Studien, 9, 351–382.

Koh, K., & Meyer, D. E. (1991). Function learning: Induction of con-tinuous stimulus-response relations. Journal of ExperimentalPsychology: Learning, Memory and Cognition, 17, 811–836.

Kohler, I. (1964). The formation and transformation of the percep-tual world. Psychological Issues, 3, 1–173.

Kornhuber, H. H., & Deecke, L. (1965). Hirnpotentialänderungenbei Willkürbewegungen und passiven Bewegungen des Men-schen: Bereitschaftspotential und reafferente Potentiale[Changes of brain potentials in voluntary and passive move-ments of man: Bereitschaftspotential and reafferente potentiale].Pflügers Archiv für die gesamte Physiologie, 284, 1–17.

Krakauer, J. W., Ghilardi, M. F., & Ghez, C. (1999). Independentlearning of internal models for kinematic and dynamic control ofreaching. Nature Neuroscience, 2, 1026–1031.

Krampe, R. T., Kliegl, R., Mayr, U., Engbert, R., & Vorberg, D.(2000). The fast and the slow of skilled bimanual rhythm produc-tion: Parallel versus integrated timing. Journal of ExperimentalPsychology: Human Perception and Performance, 26, 206–233.

Kravitz, J. H. (1972). Conditioned adaptation to prismatic displace-ment. Perception & Psychophysics, 11, 38–42.

Laabs, G. J. (1974). The effect of interpolated motor activity on theshort-term retention of movement distance and end-location.Journal of Motor Behavior, 6, 279–288.

Lajoie, Y., Paillard, J., Teasdale, N., Bard, C., Fleury, M., Forget, R.,& Lamarre, Y. (1992). Mirror drawing in a deafferented patientand normal subjects: Visuoproprioceptive conflict. Neurology,42, 1104–1106.

Lashley, K. S. (1917). The accuracy of movement in the absence ofexcitation from the moving organ. American Journal of Physiol-ogy, 43, 169–194.

Lashley, K. S. (1951). The problem of serial order in behavior. InL. A. Jeffress (Ed.), Cerebral mechanism in behavior (pp. 112–131). New York: Wiley.

Latash, M. L., & Jaric, S. (in press). The organization of drinking:Postural characteristics of the arm-head coordination. Journal ofMotor Behavior.

Lee, D. N. (1976). A theory of visual control of braking based oninformation about time-to-collision. Perception, 5, 437–459.

Lee, T. D., Blandin, Y., & Proteau, L. (1996). Effects of task in-structions and oscillation frequency on bimanual coordination.Psychological Research, 59, 100–106.

Legge, D. (1965). Analysis of visual and proprioceptive compo-nents of motor skill by means of a drug. British Journal of Psy-chology, 56, 243–254.

Lépine, D., Glencross, D., & Requin, J. (1989). Some experimentalevidence for and against a parametric conception of movementprogramming. Journal of Experimental Psychology: HumanPerception and Performance, 15, 347–362.

Lippold, O. C. J. (1952). The relation between integrated actionpotentials in a human muscle and its isometric tension. Journalof Physiology, 117, 492–499.

Lippold, O. C. J., Redfearn, J. W. T., & Vuco, j. (1960). The elec-tromyography of fatigue. Ergonomics, 3, 121–131.

Mack, A., Heuer, F., Villardi, K., & Chambers, D. (1985). Thedissociation of position and extent in Müller-Lyer figures.Perception & Psychophysics, 37, 335–344.

MacKenzie, C. L., Marteniuk, R. G., Dugas, C., Liske, D., &Eickmeier, B. (1987). Three-dimensional movement trajectoriesin Fitts’ task: Implications for control. Quarterly Journal ofExperimental Psychology, 39A, 629–647.

Marteniuk, R. G., & MacKenzie, C. L. (1980). A preliminary theoryof two-hand co-ordinated control. In G. E. Stelmach & J. Requin

References 351

(Eds.), Tutorials in motor behavior (pp. 185–197). Amsterdam:North-Holland.

Marteniuk, R. G., MacKenzie, C. L., & Baba, D. M. (1984). Biman-ual movement control: Information processing and interactioneffects. Quarterly Journal of Experimental Psychology, 36A,335–365.

Marteniuk, R. G., MacKenzie, C. L., Jeannerod, M., Athènes, S., &Dugas, C. (1987). Constraints on human arm movement trajec-tories. Canadian Journal of Psychology, 41, 365–378.

Massion, J. (1992). Movement, posture, and equilibrium: Interac-tions and coordination. Progress in Neurobiology, 38, 35–56.

McDowell, M. J., & Wolff, P. H. (1997). A functional analysis ofhuman mirror movements. Journal of Motor Behavior, 29,85–96.

McGonigle, B. O., & Flook, J. (1978). Long-term retention of singleand multistate prismatic adaptation by human. Nature, 272,364–366.

McIntyre, J., Stratta, F., & Lacquaniti, F. (1998). Short-term mem-ory for reaching to visual targets: Psychophysical evidence forbody-centered reference frames. Journal of Neuroscience, 18,8423–8435.

McLeod, P., McLaughlin, C., & Nimmo-Smith, I. (1985). Informa-tion encapsulation and automaticity: Evidence from the visualcontrol of finely timed actions. In M. I. Posner & O. S. M. Marin(Eds.), Attention and performance XI (pp. 391–406). Hillsdale,NJ: Erlbaum.

Merz, F., Kalveram, K.-T., & Huber, K. (1981). Der Einfluß kogni-tiver Faktoren auf Steuerleistungen [The role of cognitive factorsin manual tracking]. In L. Tent (Ed.), Erkennen–Wollen–Handeln (pp. 327–335). Göttingen, Germany: Hogrefe.

Metz, A. M. (1970). Änderungen der myoelektrischen Aktivitätwährend eines sensomotorischen Lernprozesses [Changes ofmyoelectric activity during motor learning]. Zeitschrift für Psy-chologie, 178, 51–88.

Meyer, D. E., Abrams, R. A., Kornblum, S., Wright, C. E., & Smith,J. E. K. (1988). Optimality in human motor performance: Idealcontrol of rapid aimed movements. Psychological Review, 95,340–370.

Meyer, D. E., Smith, J. E. K., & Wright, C. E. (1982). Models for thespeed and accuracy of aimed movements. Psychological Review,89, 449–482.

Monsell, S. (1986). Programming of complex sequences: Evidencefrom the timing of rapid speech and other productions. InH. Heuer & C. Fromm (Eds.), Generation and modulation ofaction patterns (pp. 72–86). Berlin, Germany: Springer.

Müller, H. (2001). Ausführungsvariabilität und Ergebniskonstanz[Variability of execution and constancy of outcomes].Lengerich: Pabst Science Publishers.

Müller, H., & Loosch, E. (1999). Functional variability and an equi-final path of movement during targeted throwing. Journal ofHuman Movement Studies, 36, 103–126.

Näätänen, R., & Merisalo, A. (1977). Expectancy and preparation insimple reaction time. In S. Dornic (Ed.), Attention and perfor-mance VI (pp. 115–138) Hillsdale, NJ: Erlbaum.

Nelson, W. L. (1983). Physical principles for economies of skilledmovements. Biological Cybernetics, 46, 135–147.

Newell, K. M., & van Emmerik, R. E. A. (1989). The acquisition ofcoordination: Preliminary analysis of learning to write. HumanMovement Science, 8, 17–32.

Nielsen, T. I. (1963). Volition: A new experimental approach. Scan-dinavian Journal of Psychology, 4, 225–230.

Noble, M., Fitts, P. M., & Warren, C. E. (1955). The frequencyresponse of skilled subjects in a pursuit tracking task. Journal ofExperimental Psychology, 49, 249–256.

Nougier, V., Bard, C., Fleury, M., Teasdale, N., Cole, J., Forget, R.,Paillard, J., & Lamarre, Y. (1996). Control of single-joint move-ments in deafferented patients: Evidence for amplitude codingrather than position control. Experimental Brain Research, 109,473–482.

Paillard, J. (1982). The contribution of peripheral and central visionto visually guided reaching. In D. J. Ingle, M. A. Goodale, &R. J. W. Mansfield (Eds.), Analysis of visual behavior(pp. 367–385). Cambridge, MA: MIT Press.

Peper, C. E. (1995). Tapping dynamics. Unpublished doctoral dis-sertation, Free University Amsterdam.

Peper, C. E., & Beek, P. J. (1998). Are frequency-induced transitionsin rhythmic coordination mediated by a drop in amplitude?Biological Cybernetics, 79, 291–300.

Peper, L., Bootsma, R. J., Mestre, D. R., & Bakker, F. C. (1994).Catching balls: How to get the hand to the right place at the righttime. Journal of Experimental Psychology: Human Perceptionand Performance, 20, 591–612.

Perenin, M. T., & Jeannerod, M. (1978). Visual function within thehemianopic field following early cerebral hemidecortication inman: I. Spatial localization. Neuropsychologia, 16, 1–13.

Peters, M. (1981). Attentional asymmetries during concurrentbimanual performance. Quarterly Journal of Experimental Psy-chology, 33A, 95–103.

Pew, R.W. (1966). Acquisition of hierarchical control over the tem-poral organization of skill. Journal of Experimental Psychology,71, 764–771.

Plamondon, R., & Alimi, A. M. (1997). Speed/accuracy trade-offs intarget-directed movements. Behavioral and Brain Sciences, 20,279–349.

Polit, A., & Bizzi, E. (1979). Characteristics of motor programsunderlying arm movements in monkeys. Journal of Neurophysi-ology, 42, 183–194.

Poulton, E. C. (1957). On the stimulus and response in pursuit track-ing. Journal of Experimental Psychology, 53, 189–194.

Preilowski, B. (1972). Possible contribution of the anterior fore-brain commissures to bilateral motor coordination. Neuropsy-chologia, 10, 267–277.

352 Motor Control

Prinz, W. (1992). Why don’t we perceive our brain states? EuropeanJournal of Cognitive Psychology, 4, 1–20.

Rack, P. M. H., & Westbury, D. R. (1969). The effects of length andstimulus rate on tension in the isometric cat soleus muscle.Journal of Physiology, 204, 443–460.

Rinkenauer, G., Ulrich, R., & Wing, A. (2001). Brief bimanual forcephases: Correlations between the hands in force and time. Jour-nal of Experimental Psychology: General, Human Perceptionand Performance, 27, 1485–1497.

Rosenbaum, D. A. (1980). Human movement initiation: Specifica-tion of arm, direction, and extent. Journal of Experimental Psy-chology: General, 109, 444–474.

Rosenbaum, D. A. (1983). The movement precuing technique:Assumptions, applications, and extensions. In R. A. Magill (Ed.),Memory and control of action (pp. 231–274). Amsterdam: North-Holland.

Rosenbaum, D. A., & Jorgensen, M. J. (1992). Planning macro-scopic aspects of manual control. Human Movement Science, 11,61–69.

Rosenbaum, D. A., & Krist, H. (1996). Antecedents of action. InH. Heuer & S. W. Keele (Eds.), Handbook of perception andaction: Vol. 2. Motor skills (pp. 3–69). London: Academic Press.

Rosenbaum, D. A., Loukopoulos, L. D., Meulenbroek, R. G. J.,Vaughan, J., & Engelbrecht, S. E. (1995). Planning reaches byevaluating stored postures. Psychological Review, 102, 28–67.

Rosenbaum, D. A., Slotta, J. D., Vaughan, J., & Plamondon, R.(1991). Optimal movement selection. Psychological Science, 2,86–91.

Rosenbaum, D. A., Vaughan, J., Barnes, H. J., & Jorgensen, M. J.(1992). Time course of movement planning: Selection of hand-grips for object manipulation. Journal of Experimental Psychol-ogy: Learning, Memory, and Cognition, 18, 1058–1073.

Roth, K. (1988). Investigations on the basis of the generalized motorprogramme hypothesis. In O. G. Meijer & K. Roth (Eds.),Complex movement behaviour: “The” motor-action controversy(pp. 261–288). Amsterdam: North-Holland.

Ruitenbeek, J. C. (1984). Invariants in loaded goal directed move-ments. Biological Cybernetics, 51, 11–20.

Saltzman, E., & Kelso, J. A. S. (1987). Skilled actions: A task-dynamic approach. Psychological Review, 94, 84–106.

Sangals, J. (1997). Der Einfluß der Bewegungsrückmeldung auf dasErlernen nichtlinearer Werkzeugtransformationen [The effect ofmovement feedback on the learning of nonlinear transforma-tions]. Unpublished doctoral dissertation, Philipps-Universitat,Marburg, Marburg, Germany.

Sangals, J., Heuer, H., Manzey, D., & Lorenz, B. (1999). Changedvisuomotor transformations during and after prolonged micro-gravity. Experimental Brain Research, 129, 378–390.

Savelsbergh, G. J. P., Whiting, H. T. A., & Bootsma, R. J. (1991).Grasping tau. Journal of Experimental Psychology: HumanPerception and Performance, 17, 315–322.

Schiff, W., & Detwiler, M. L. (1979). Information used in judgingimpending collision. Perception, 8, 647–658.

Schmidt, R. A. (1972). The Index of Preprogramming (IP): A statis-tical method for evaluating the role of feedback in simple move-ments. Psychonomic Science, 27, 83–85.

Schmidt, R. A. (1975). A schema theory of discrete motor skilllearning. Psychological Review, 82, 225–260.

Schmidt, R. A. (1980). On the theoretical status of time in motor-program representations. In G. E. Stelmach & J. Requin (Eds.),Tutorials in motor behavior (pp. 145–165). Amsterdam: North-Holland.

Schmidt, R. A. (1985). The search for invariance in skilled move-ment behavior. Research Quarterly for Exercise and Sport, 56,188–200.

Schmidt, R. A. (1989). Unintended acceleration: A review of humanfactors contributions. Human Factors, 31, 345–364.

Schmidt, R. A., & McGown, C. (1980). Terminal accuracy of unex-pectedly loaded rapid movements: Evidence for a mass-springmechanism in programming. Journal of Motor Behavior, 12,149–161.

Schmidt, R. A., & Russell, D. G. (1972). Movement velocity andmovement time as determinants of degree of preprogramming insimple movements. Journal of Experimental Psychology, 96,315–320.

Schmidt, R. A., Sherwood, D. E., Zelaznik, H. N., & Leikind, B. J.(1985). Speed-accuracy tradeoffs in motor behavior: Theories ofimpulse variability. In H. Heuer, U. Kleinbeck, & K.-H. Schmidt(Eds.), Motor behavior: Programming, control, and acquisition(pp. 79–123). Berlin, Germany: Springer.

Schmidt, R. A., Zelaznik, H. N., Hawkins, B., Frank, J. S., & Quinn,J. T. (1979). Motor-output variability: A theory for the accuracyof rapid motor acts. Psychological Review, 86, 415–451.

Schöner, G. (1994). From interlimb coordination to trajectoryformation: Common dynamical principles. In S. P. Swinnen,H. Heuer, J. Massion, & P. Casaer (Eds.), Interlimb coordina-tion. Neural, dynamical, and cognitive constraints (pp. 339–368). San Diego, CA: Academic Press.

Schöner, G., & Kelso, J. A. S. (1988). A synergetic theory ofenvironmentally-specified and learned patterns of movement co-ordination. I: Relative phase dynamics. Biological Cybernetics,58, 71–80.

Schott, G. D., & Wyke, M. A. (1981). Congenital mirror move-ments. Journal of Neurology, Neurosurgery, & Psychiatry, 44,586–599.

Schouten, J. F., & Becker, J. A. M. (1967). Reaction time and accu-racy. Acta Psychologica, 27, 143–153.

Serrien, D. J., Bogaerts, H., Suy, E., & Swinnen, S.P. (1999). Theidentification of coordination constraints across planes of mo-tion. Experimental Brain Research, 128, 250–255.

Shadmehr, R., & Brashers-Krug, T. (1997). Functional stages in theformation of human long-term motor memory. Journal of Neu-roscience, 17, 409–419.

References 353

Shadmehr, R., & Moussavi, Z. M. K. (2000). Spatial generalizationfrom learning dynamics of reaching movements. Journal ofNeuroscience, 20, 7807–7815.

Shadmehr, R., & Mussa-Ivaldi, F. A. (1994). Adaptive representa-tion of dynamics during learning of a motor task. Journal ofNeuroscience, 14, 3208–3224.

Sherwood, D. E. (1991). Distance and location assimilation in rapidbimanual movement. Research Quarterly for Exercise andSport, 62, 302–308.

Smeets, J. B. J., & Brenner, E. (1995). Perception and action arebased on the same visual information: Distinction between posi-tion and velocity. Journal of Experimental Psychology: HumanPerception and Performance, 21, 19–31.

Smeets, J. B. J., Erkelens, C. J., & Denier van der Gon, J. J. (1995).Perturbations of fast goal-directed arm movements: Differentbehavior of early and late EMG responses. Journal of MotorBehavior, 27, 77–88.

Spijkers, W. (1993). Sehen und Handeln: Die Rolle visueller Infor-mation bei zielgerichteten Bewegungen [Perception and action:The role of visual information in aimed movements]. Aachen,Germany: Shaker.

Spijkers, W., & Heuer, H. (1995). Structural constraints on the per-formance of symmetrical bimanual movements with differentamplitudes. Quarterly Journal of Experimental Psychology, 48A,716–740.

Spijkers, W., Heuer, H., Kleinsorge, T., & Steglich, C. (2000). Thespecification of movement amplitudes for the left and right hand:Evidence for transient parametric coupling from overlapping-task performance. Journal of Experimental Psychology: HumanPerception and Performance, 26, 1091–1105.

Spijkers, W., Heuer, H., Kleinsorge, T., & van der Loo, H. (1997).Preparation of bimanual movements with same and different am-plitudes: Specification interference as revealed by reaction time.Acta Psychologica, 96, 207–227.

Spijkers, W., Tachmatzidis, K., Debus, G., Fischer, M., & Kausche,I. (1994). Temporal coordination of alternative and simultaneousaiming movements of constrained timing structure. Psychologi-cal Research, 57, 20–29.

Steglich, C., Heuer, H., Spijkers, W., & Kleinsorge, T. (1999).Bimanual coupling during the specification of isometric forces.Experimental Brain Research, 129, 302–316.

Stelmach, G. E. (1982). Motor control and motor learning:The closed-loop perspective. In J.A. S. Kelso (Ed.), Human motorbehavior: An introduction (pp. 93–115). Hillsdale, NJ: Erlbaum.

Stelmach, G. E., Kelso, J. A. S., & Wallace, S. A. (1975). Preselec-tion in short-term motor memory. Journal of Experimental Psy-chology: Human Learning and Memory, 1, 745–755.

Stimpel, E. (1933). Der Wurf [The throw]. Neue PsychologischeStudien, 9, 105–138.

Stratton, G. M. (1896). Some preliminary experiments in visionwithout inversion of the retinal image. Psychological Review, 3,611–617.

Stratton, G. M. (1897a). Upright vision and the retinal image. Psy-chological Review, 4, 182–187.

Stratton, G. M. (1897b). Vision without inversion of the retinalimage. Psychological Review, 4, 341–360, 463–481.

Stucchi, N., & Viviani, P. (1993). Cerebral dominance and asyn-chrony between bimanual two-dimensional movements. Journalof Experimental Psychology: Human Perception and Perfor-mance, 19, 1200–1220.

Summers, J. J., Rosenbaum, D. A., Burns, B. D., & Ford, S. K.(1993). Production of polyrhythms. Journal of Experimen-tal Psychology: Human Perception and Performance, 19, 416–428.

Swinnen, S. P., Jardin, K., & Meulenbroek, R. (1996). Between-limb asynchronies during bimanual coordination: Effects ofmanual dominance and attentional cueing. Neuropsychologica,34, 1203–1213.

Taub, E., Goldberg, I. A., & Taub, P. (1975). Deafferentation inmonkeys: Pointing at a target without visual feedback. Experi-mental Neurology, 46, 178–186.

Teasdale, N., Forget, R., Bard, C., Paillard, J., Fleury, M., &Lamarre, Y. (1993). The role of proprioceptive information forthe production of isometric forces and for handwriting tasks.Acta Psychologica, 82, 179–191.

Tendick, F., Jennings, R. W., Tharp, G., & Stark, L. (1993). Sensingand manipulation problems in endoscopic surgery: experiment,analysis, and observation. Presence, 2, 66–81.

Teulings, H.-L. (1996). Handwriting movement control. In H.Heuer & S. W. Keele (Eds.), Handbook of perception and ac-tion: Vol. 2. Motor skills (pp. 561–613). London: AcademicPress.

Todor, J. I., & Lazarus, J. C. (1986). Exertion level and the intensityof associated movements. Developmental Medicine & ChildNeurology, 28, 205–212.

Tresilian, J. R. (1999). Analysis of recent empirical challenges to anaccount of interceptive timing. Perception & Psychophysics, 61,515–528.

Tuller, B., & Kelso, J. A. S. (1989). Environmentally-specified pat-terns of movement coordination in normal and split-brain sub-jects. Experimental Brain Research, 75, 306–316.

Tyldesley, D. A., & Whiting, H. T. A. (1975). Operational timing.Journal of Human Movement Studies, 1, 172–177.

Ungerleider, L. G., & Mishkin, M. (1982). Two cortical visual sys-tems. In D. J. Ingle, M. A. Goodale, & R. J. W. Mansfield (Eds.),Analysis of visual behavior (pp. 549–586). Cambridge, MA: MITPress.

Uno, Y., Kawato, M., & Suzuki, R. (1989). Formation and control ofoptimal trajectory in human multijoint arm movement: Mini-mum torque-change model. Biological Cybernetics, 61, 89–101.

Vindras, P., & Viviani, P. (1998). Frames of reference and controlparameters in visuo manual pointint. Journal of ExperimentalPsychology: Human Perception and Performance, 24, 569–591.

354 Motor Control

Virjii-Babul, N., Cooke, J. D., & Brown, S. H. (1994). Effects ofgravitational forces on single joint arm movements in humans.Experimental Brain Research, 99, 338–346.

Viviani, P., & Flash, T. (1995). Minimum-jerk, two-thirds powerlaw, and isochrony: Converging approaches to movement plan-ning. Journal of Experimental Psychology: Human Perceptionand Performance, 21, 32–53.

Viviani, P., Perani, D., Grassi, F., Bettinardi, V., & Fazio, F. (1998).Hemispheric asymmetries and bimanual asynchrony in left- andright-handers. Experimental Brain Research, 120, 531–536.

Voigt, E. (1933). Über den Aufbau von Bewegungsgestalten [On thecomposition of movement Gestalts]. Neue Psychologische Stu-dien, 9, 1–32.

Vorberg, D., & Wing, A. M. (1996). Modeling variability anddependence in timing. In H. Heuer & S. W. Keele (Eds.), Hand-book of perception and action: Vol. 2. Motor skills (pp. 181–262). London: Academic Press.

Wadman, W. J., Denier van der Gon, J. J., Geuze, R. H., & Mol, C.R. (1979). Control of fast goal-directed arm movements. Journalof Human Movement Studies, 5, 3–17.

Welch, R. B. (1971). Discriminative conditioning of prism adapta-tion. Perception & Psychophysics, 10, 90–92.

Welch, R. B. (1972). The effect of experienced limb identity uponadaptation to simulated displacement of the visual field. Percep-tion & Psychophysics, 12, 453–456.

Welch, R. B. (1978). Perceptual modification. Adapting to alteredsensory environments. New York: Academic Press.

Welch, R. B., Bridgeman, B., Anand, S., & Browman, K. E. (1993).Alternating prism exposure causes dual adaptation and general-ization to a novel displacement. Perception & Psychophysics,54, 195–204.

Whiting, H. T. A., & Cockerill, I. M. (1974). Eyes on hand—eyes ontarget? Journal of Motor Behavior, 6, 27–32.

Winter, D. A., & Robertson, D. G. E. (1978). Joint torque and energypatterns in normal gait. Biological Cybernetics, 29, 137–142.

Woodworth, R. S. (1899). The accuracy of voluntary movement.Psychological Review, 3 (Monograph Suppl. 3).

Wright, C. E., & Meyer, D. E. (1983). Conditions for a linear speed-accuracy trade-off in aimed movements. Quarterly Journal ofExperimental Psychology, 35A, 279–296.

Wulf, G., Höß, M., & Prinz, W. (1998). Instructions for motor learn-ing: differential effects of internal vs. external focus of attention.Journal of Motor Behavior, 30, 169–179.

Wulf, G., & Prinz, W. (2001). Directing attention to movementeffects enhances learning: A review. Psychonomic Bulletin andReview, 8, 648–660.

Yamanishi, J., Kawato, M., & Suzuki, R. (1980). Two coupled os-cillators as a model of the coordinated finger tapping by bothhands. Biological Cybernetics, 37, 219–225.

Young, L. R. (1969). On adaptive manual control. Ergonomics, 12,635–675.

Zattara, M., & Bouisset, S. (1986). Chronometric analysis of theposturo-kinetic programming of voluntary movement. Journalof Motor Behavior, 18, 215–223.

Zelaznik, H. N., Hawkins, B., & Kisselburgh, L. (1983). Rapid vi-sual feedback processing in single-aiming movements. Journalof Motor Behavior, 15, 217–236.

Zelaznik, H. N., Mone, S., McCabe, G. P., & Thaman, C. (1988).Role of temporal and spatial precision in determining the natureof the speed-accuracy trade-off in aimed hand movements. Jour-nal of Experimental Psychology: Human Perception and Perfor-mance, 14, 221–230.

Zelaznik, H. N., Shapiro, D. C., & Carter, M. C. (1982). Thespecification of digit and duration during motor programming: Anew method of precueing. Journal of Motor Behavior, 14,57–68.